Single-hole pressure regulation method, device and equipment for water curtain test of underground water-sealed warehouse and storage medium

Abstract

This application discloses a method, apparatus, equipment, and storage medium for single-hole pressure regulation in underground water-sealed cavern water curtain tests, relating to the field of geological engineering technology. The method includes: first, acquiring the acoustic impedance spectrum of the rock mass and performing intrinsic mode decomposition to extract geological feature components; then, using a pressure-opening coupling model based on acoustic feature extraction, temporal dynamic analysis, and physical constraint decoding to process the target pressure value, current pressure value, and geological feature components to obtain a reference valve opening; next, adjusting the water curtain orifice valve according to the reference valve opening and monitoring the pressure change acceleration in real time; calculating the opening compensation increment and damping attenuation coefficient based on the pressure change acceleration to generate the target valve opening; finally, adjusting the valve to the target opening so that the water curtain orifice reaches the target pressure value. This application enables high-precision adaptive pressure regulation of a single water curtain orifice in underground water-sealed cavern water curtain tests.

Description

Technical Field

[0001] This application relates to the field of geological engineering technology, and in particular to a single-hole pressure regulation method, device, equipment and storage medium for water curtain tests in underground water-sealed caverns. Background Technology

[0002] During the testing phase, the water curtain system is required to maintain the pressure of a single hole within the design value of ±0.05MPa to ensure that the hydraulic seal of the rock mass fissures does not leak or split. However, the seepage of the rock mass is affected by multiple sources of disturbance, such as groundwater pulsation, temperature drift and construction disturbance, and traditional manual or simple automatic adjustment is difficult to track accurately in real time.

[0003] Currently, two common solutions are "manual needle valve + pressure gauge" or "electric regulating valve + single-loop PID". The former relies on experience to approximate the target pressure step by step, while the latter uses a pressure sensor to provide real-time feedback on the deviation and uses a PID algorithm to drive the valve opening to achieve closed-loop pressure stabilization.

[0004] Existing practices have several problems: open-loop regulation suffers from setpoint drift and repeated overshoot due to valve dead zones, hysteresis, and nonlinear flow characteristics; PID closed-loop regulation only compensates for historical errors with lag and cannot predict or suppress feedforward disturbances such as groundwater pulsations; mechanical clearances, execution delays, and abrupt dead-zone switching in electric valves further amplify control distortion, making it difficult to stabilize the pressure of a single orifice within ±0.05 MPa. Therefore, achieving high-precision adaptive regulation of the pressure of a single water curtain orifice in underground water-sealed cavern water curtain tests is an urgent problem to be solved.

[0005] The above content is only used to help understand the technical solution of this application and does not represent an admission that the above content is prior art. Summary of the Invention

[0006] The purpose of this application is to provide a method, device, equipment and storage medium for single-hole pressure regulation in underground water-sealed cavern water curtain tests, aiming to solve the technical problem of how to achieve high-precision adaptive regulation of single water curtain hole pressure in underground water-sealed cavern water curtain tests.

[0007] To achieve the above objectives, this application proposes a single-hole pressure regulation method for a water curtain test in an underground water-sealed cavern, the method comprising:

[0008] Acoustic impedance spectra of the rock mass were collected, and intrinsic mode decomposition was performed on the rock mass acoustic impedance spectra to extract geological feature components.

[0009] The target pressure value, the current pressure value of the water curtain orifice, and the geological feature components of the water curtain test are processed by a pressure-opening coupling model to obtain the reference valve opening. The pressure-opening coupling model is constructed based on the acoustic feature extraction module, the temporal dynamic module, and the physical constraint decoding module.

[0010] The valve of the water curtain orifice is adjusted according to the reference valve opening, and the pressure change acceleration of the water curtain orifice is monitored during the adjustment process;

[0011] Calculate the opening compensation increment and damping attenuation coefficient based on the pressure change acceleration;

[0012] The target valve opening is generated based on the reference valve opening, the opening compensation increment, and the damping attenuation coefficient.

[0013] Adjust the valve of the water curtain orifice to the target valve opening so that the water curtain orifice reaches the target pressure value.

[0014] In one embodiment, the step of calculating the opening compensation increment and damping attenuation coefficient based on the pressure change acceleration includes:

[0015] When the pressure change acceleration is greater than a preset pressure acceleration threshold, pressure time series data is acquired, and a pressure deviation time series signal is generated based on the pressure time series data and the target pressure value.

[0016] The pressure deviation time-series signal is subjected to Hilbert-Huang transform to obtain the hydraulic transient component and the mechanical response hysteresis component;

[0017] The lag time constant of the mechanical response lag component is calculated by shock response filtering;

[0018] The valve opening compensation increment is calculated based on the valve's maximum regulating rate, the current pressure deviation, and the hysteresis time constant.

[0019] Calculate the oscillation energy density of the hydraulic transient component, and generate a damping attenuation coefficient based on the oscillation energy density.

[0020] In one embodiment, the steps for constructing the pressure-opening coupling model include:

[0021] An acoustic feature extraction module is constructed based on a one-dimensional convolutional network, an attention gating layer, a dimensionality reduction processing layer, and a regression output layer.

[0022] A temporal dynamic module is constructed based on a phase compensator, a three-dimensional convolutional layer, a time-frequency attention gating mechanism, and a bidirectional long short-term memory network.

[0023] A physical constraint decoding module is constructed based on the physical channel calculation unit, the residual channel calculation unit, the Sigmoid activation layer, and the mechanical limit protection layer.

[0024] By combining the acoustic feature extraction module, the temporal dynamics module, and the physical constraint decoding module, a pressure-opening coupling model is obtained.

[0025] In one embodiment, the pressure-opening coupling model includes an acoustic feature extraction module, a temporal dynamics module, and a physical constraint decoding module;

[0026] The step of processing the target pressure value of the water curtain test, the current pressure value of the water curtain orifice, and the geological feature components through a pressure-opening coupling model to obtain the reference valve opening includes:

[0027] The geological feature components are multi-scale convolutional encoded by the acoustic feature extraction module to obtain hydraulic conduction correction coefficients.

[0028] The time-series dynamic module extracts spatiotemporal features from historical pressure sequences and historical valve action sequences to obtain dynamic compensation vectors.

[0029] The physical constraint decoding module performs dual-channel residual fusion calculation on the hydraulic transmission correction coefficient, the dynamic compensation vector, the target pressure value of the water curtain test, and the current pressure value of the water curtain orifice to obtain the reference valve opening.

[0030] In one embodiment, the acoustic feature extraction module includes a one-dimensional convolutional network, an attention gating layer, a dimensionality reduction processing layer, and a regression output layer. The one-dimensional convolutional network includes a first convolutional layer, a second convolutional layer, and a third convolutional layer. The kernel length of the second convolutional layer is greater than the kernel length of the first convolutional layer, and the kernel length of the third convolutional layer is greater than the kernel length of the second convolutional layer.

[0031] The step of performing multi-scale convolutional encoding on the geological feature components through the acoustic feature extraction module to obtain hydraulic conduction correction coefficients includes:

[0032] The geological feature components are subjected to high-frequency feature extraction through the first convolutional layer to obtain the microfracture feature vector;

[0033] The geological feature components are extracted using the second convolutional layer to obtain the interface effect feature vector.

[0034] The permeability tensor is obtained by extracting low-frequency features from the geological feature components through the third convolutional layer.

[0035] The attention gating layer is used to weight and fuse the microcrack feature vector, the interface effect feature vector, and the permeability tensor to obtain a fused feature tensor.

[0036] The fusion feature tensor is channel-compressed through the dimensionality reduction layer to obtain a low-dimensional feature vector;

[0037] The hydraulic conduction correction coefficient is obtained by performing a linear transformation on the low-dimensional feature vector through the regression output layer.

[0038] In one embodiment, the temporal dynamic module includes a phase compensator, a first bidirectional long short-term memory network, a three-dimensional convolutional layer, a time-frequency attention gating mechanism, and a second bidirectional long short-term memory network;

[0039] The step of extracting spatiotemporal features from historical pressure sequences and historical valve action sequences using the time-series dynamic module to obtain dynamic compensation vectors includes:

[0040] The phase compensator is used to perform sensor delay correction on the historical pressure sequence to obtain the phase-corrected pressure sequence.

[0041] The phase-corrected stress sequence is subjected to temporal feature extraction through the first bidirectional long short-term memory network to obtain a temporal stress feature vector. The first bidirectional long short-term memory network is configured with hidden units of a first preset time step and a first preset dimension.

[0042] The historical valve action sequence is spatially encoded by the three-dimensional convolutional layer to obtain the spatiotemporal features of valve action;

[0043] The time-frequency attention gating mechanism is used to weight and fuse the temporal pressure feature vector and the spatiotemporal features of valve action to obtain pressure-valve action fusion features.

[0044] The second bidirectional long short-term memory network is used to extract decision features from the pressure-valve action fusion features to obtain a dynamic compensation vector. The second bidirectional long short-term memory network is configured with a second preset time step, a second preset dimension of hidden units, and an acoustic gating mechanism. The second preset time step is smaller than the first preset time step, and the second preset dimension is smaller than the first preset dimension.

[0045] In one embodiment, the physical constraint decoding module includes a physical channel calculation unit, a residual channel calculation unit, a Sigmoid activation layer, and a mechanical limit protection layer;

[0046] The step of obtaining the reference valve opening by performing dual-channel residual fusion calculation on the hydraulic conduction correction coefficient, the dynamic compensation vector, the target pressure value of the water curtain test, and the current pressure value of the water curtain orifice through the physical constraint decoding module includes:

[0047] The physical channel calculation unit performs a nonlinear transformation on the pressure difference between the target pressure value of the water curtain test and the current pressure value of the water curtain orifice to obtain the basic opening component.

[0048] The dynamic compensation opening component is obtained by mapping the dynamic compensation vector and the pressure difference value through the residual channel calculation unit.

[0049] The dynamic compensation aperture component is subjected to amplitude limiting processing through the Sigmoid activation layer to obtain the compressed compensation component;

[0050] The reference valve opening is obtained by superimposing the basic opening component and the compression compensation component, and constraining the superposition result through the mechanical limit protection layer.

[0051] Furthermore, to achieve the above objectives, this application also proposes a single-hole pressure regulating device for underground water-sealed cavern water curtain tests, the device comprising:

[0052] The mode decomposition module is used to acquire the acoustic impedance spectrum of the rock mass and perform intrinsic mode decomposition on the acoustic impedance spectrum of the rock mass to extract geological feature components.

[0053] The model processing module is used to process the target pressure value, the current pressure value of the water curtain orifice, and the geological feature components of the water curtain test through a pressure-opening coupling model to obtain the reference valve opening. The pressure-opening coupling model is constructed based on the acoustic feature extraction module, the temporal dynamic module, and the physical constraint decoding module.

[0054] The first valve adjustment module is used to adjust the valve of the water curtain orifice according to the reference valve opening, and to monitor the pressure change acceleration of the water curtain orifice during the adjustment process;

[0055] The compensation increment calculation module is used to calculate the opening compensation increment and damping attenuation coefficient based on the pressure change acceleration.

[0056] The valve opening calculation module is used to generate the target valve opening based on the reference valve opening, the opening compensation increment, and the damping attenuation coefficient.

[0057] The second valve adjustment module is used to adjust the valve of the water curtain orifice to the target valve opening so that the water curtain orifice reaches the target pressure value.

[0058] Furthermore, to achieve the above objectives, this application also proposes a single-hole pressure regulating device for a water curtain test in an underground water-sealed cavern. The device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor. The computer program is configured to implement the steps of the single-hole pressure regulating method for the water curtain test in an underground water-sealed cavern as described above.

[0059] In addition, to achieve the above objectives, this application also proposes a storage medium, which is a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements the steps of the single-hole pressure regulation method for the underground water-sealed cavern water curtain test as described above.

[0060] In addition, to achieve the above objectives, this application also provides a computer program product, which includes a computer program that, when executed by a processor, implements the steps of the single-hole pressure regulation method for the underground water-sealed cavern water curtain test as described above.

[0061] One or more technical solutions proposed in this application have at least the following technical effects:

[0062] First, the acoustic impedance spectrum of the rock mass is collected and intrinsic mode decomposition is performed to extract geological feature components. This step enables real-time perception of the impact of internal rock fissures and water content changes on water seal performance, providing a geological basis for subsequent precise regulation. Next, a pressure-opening coupling model is used to process the target pressure value, current pressure value, and geological feature components to obtain the baseline valve opening. This model integrates acoustic feature extraction, temporal dynamic analysis, and physical constraint decoding, comprehensively considering multiple factors affecting the initial adjustment requirements of the valve opening. Then, the water curtain orifice valve is adjusted according to the baseline valve opening, and the acceleration of pressure changes is monitored. This step captures the intensity of pressure fluctuations in real time, providing immediate feedback for subsequent compensation. The opening compensation increment and damping attenuation coefficient are calculated based on the pressure change acceleration. This allows for dynamic adjustment of the valve opening while suppressing hydraulic oscillations, ensuring the stability of pressure regulation. Finally, the target valve opening is generated based on the baseline valve opening, the opening compensation increment, and the damping attenuation coefficient, and the valve is adjusted to this opening, enabling the water curtain orifice to reach the target pressure value. This series of steps enables high-precision adaptive adjustment of the pressure in a single water curtain hole during underground water-sealed cavern water curtain tests, significantly improving test efficiency and safety. Attached Figure Description

[0063] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0064] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0065] Figure 1This is a flowchart illustrating the single-hole pressure regulation method for the water curtain test of the underground water-sealed cavern in this application.

[0066] Figure 2 This is a schematic diagram of the module structure of the pressure-opening coupling model provided in Embodiment 1 of the single-hole pressure regulation method for the underground water-sealed cavern water curtain test of this application;

[0067] Figure 3 This is a flowchart illustrating Example 2 of the single-hole pressure regulation method for the underground water-sealed cavern water curtain test in this application.

[0068] Figure 4 A time-series flowchart illustrating the single-hole pressure regulation method for the underground water-sealed cavern water curtain test provided in Embodiment 2 of this application;

[0069] Figure 5 This is a schematic diagram of the module structure of the single-hole pressure regulating device for the underground water-sealed cavern water curtain test according to an embodiment of this application;

[0070] Figure 6 This is a schematic diagram of the equipment structure of the hardware operating environment involved in the single-hole pressure regulation method of the underground water-sealed cavern water curtain test in the embodiments of this application.

[0071] The purpose, features, and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0072] It should be understood that the specific embodiments described herein are merely illustrative of the technical solutions of this application and are not intended to limit this application.

[0073] To better understand the technical solution of this application, a detailed description will be provided below in conjunction with the accompanying drawings and specific implementation methods.

[0074] It should be noted that the executing entity of this application embodiment can be a computing service device with data processing, network communication, and program execution functions, such as a tablet computer, personal computer, or mobile phone, or an electronic device capable of realizing the above functions, such as a water curtain hole dynamic adjustment system. The following uses a water curtain hole dynamic adjustment system as an example to describe this embodiment and the following embodiments.

[0075] Based on this, the embodiments of this application provide a single-hole pressure adjustment method for a water curtain test in an underground water-sealed cavern, referring to... Figure 1 , Figure 1 This is a schematic flowchart of the first embodiment of the single-hole pressure regulation method for the underground water-sealed cavern water curtain test of this application.

[0076] In this embodiment, the single-hole pressure adjustment method for the underground water-sealed cavern water curtain test includes steps S10~S60:

[0077] Step S10: Collect the acoustic impedance spectrum of the rock mass and perform intrinsic mode decomposition on the acoustic impedance spectrum of the rock mass to extract geological feature components.

[0078] It should be noted that the rock mass acoustic impedance spectrum refers to the characteristic curve showing the continuous change of acoustic impedance with frequency under acoustic excitation. It comprehensively reflects the influence of rock mass density, elastic modulus, and internal pore and fracture structure on the propagation of sound waves at different frequencies. Geological characteristic components refer to independent signal components corresponding to specific geological structures or differences in physical properties extracted from the rock mass acoustic impedance spectrum through mathematical decomposition methods. These components can directly map essential geological information such as bedding, fractures, water saturation, or lithological changes in the rock mass.

[0079] Understandably, the water curtain hole dynamic adjustment system first transmits pulsed sound waves in 0.5 kHz steps, sweeping from 1 kHz to 100 kHz using a broadband piezoelectric transducer installed in the test hole, and simultaneously records the echo amplitude and phase. The measured data is then converted into an acoustic impedance-frequency curve in real time to obtain complete spectral information covering the dominant frequency response of the fracture. Next, the system performs intrinsic mode decomposition on this curve. First, it calculates the envelope mean using cubic splines and iteratively selects 6–8 IMFs (Intrinsic Mode Functions). Then, it calculates the energy percentage and center frequency of each IMF. Based on a preset threshold, components with energy greater than 5% and center frequencies falling within the 10–50 kHz range of the fracture sensitive zone are marked as geological characteristic components. This aims to separate signals directly related to the permeability of the surrounding rock. Subsequently, the system maps the energy weights of these components to correction coefficients between 0 and 1, corrects the valve opening, and adjusts the drive current of the electric needle valve in real time, ensuring that the single-hole pressure remains within the target value ±0.05 MPa even when encountering sudden changes in rock seepage.

[0080] Step S20: The target pressure value of the water curtain test, the current pressure value of the water curtain orifice, and the geological feature components are processed by the pressure-opening coupling model to obtain the reference valve opening. The pressure-opening coupling model is constructed based on the acoustic feature extraction module, the temporal dynamic module, and the physical constraint decoding module.

[0081] It should be noted that the pressure-opening coupling model refers to a mathematical framework that unifies the nonlinear mapping relationship between the water curtain orifice pressure and the valve opening, rock mass acoustic feedback, and engineering safety boundaries. This framework is used to convert target pressure, measured pressure, and geological disturbances into valve opening. The target pressure value refers to the pressure stabilization setpoint that the water curtain orifice must reach within a specific time period, as required by the test scheme, and is given by the cavern design water seal criteria. The current pressure value refers to the actual water pressure inside the water curtain orifice measured in real time by the sensor, representing the immediate feedback quantity of the closed-loop regulation. The reference valve opening refers to the percentage of the basic valve stroke output after model calculation, without considering additional disturbance compensation, serving as the initial value for subsequent fine-tuning. The acoustic feature extraction module refers to a rock mass acoustic signal processor based on a convolutional neural network. Its function is to decompose the acoustic impedance spectrum into three geological feature components: microfracture density, rock layer interface permeability, and macroscopic anisotropy coefficient, outputting a hydraulic conduction correction coefficient γ. The temporal dynamic module refers to a spatiotemporal feature processor that integrates a long short-term memory network and three-dimensional convolution. It generates a dynamic compensation vector δ by analyzing the evolution trend of historical pressure sequences and the spatial distribution of valve actions to correct equipment response lag. The physical constraint decoding module refers to a control command generator that embeds fluid dynamics equations and mechanical limit rules. It calculates the opening value through dual-channel residual fusion (physical channel + compensation channel) and applies 95% / 5% mechanical limit protection to ensure that the output command complies with engineering safety constraints.

[0082] Understandably, the system first packages the target pressure value given by the test plan, the current pressure value returned by the sensor in real time, and three geological feature components output by the acoustic feature extraction module—microfracture density, rock interface permeability, and macroscopic anisotropy coefficient—and sends them into the pressure-opening coupling model. Inside the model, the hydraulic transmission correction coefficient γ is first calculated using the acoustic components. Then, the time-series dynamic module reads the pressure-valve action sequence of the past 30 seconds to generate a dynamic compensation vector δ. Subsequently, the physical constraint decoding module substitutes γ, δ, and the target-current pressure difference into the fluid dynamics equation. The valve stroke percentage is calculated through dual-channel residual fusion, and 95% / 5% mechanical limit protection is automatically applied. Finally, this percentage without disturbance compensation is output as the reference valve opening.

[0083] As an example, the construction steps of the pressure-opening coupling model include: constructing an acoustic feature extraction module based on a one-dimensional convolutional network, an attention gating layer, a dimensionality reduction processing layer, and a regression output layer; constructing a temporal dynamic module based on a phase compensator, a three-dimensional convolutional layer, a time-frequency attention gating mechanism, and a bidirectional long short-term memory network; constructing a physical constraint decoding module based on a physical channel computation unit, a residual channel computation unit, a sigmoid activation layer, and a mechanical constraint protection layer; and combining the acoustic feature extraction module, the temporal dynamic module, and the physical constraint decoding module to obtain the pressure-opening coupling model.

[0084] A one-dimensional convolutional network is a neural network layer that performs local feature scanning along the frequency axis on the input one-dimensional acoustic impedance spectrum, used to extract frequency domain patterns sensitive to fractures. An attention-gated layer is a gating mechanism that automatically amplifies key frequency features and suppresses redundant information through learnable weights. A dimensionality reduction layer is a linear mapping layer that compresses high-dimensional convolutional features into compact low-dimensional vectors to reduce subsequent computation. A regression output layer is a fully connected layer that directly maps the dimensionality-reduced features to the numerical values ​​of three geological feature components.

[0085] A phase compensator is a filtering unit that first performs phase alignment on the pressure-valve time series to eliminate misalignment caused by sampling time differences. A 3D convolutional layer is a computational layer that slides convolution kernels across a 3D tensor of time-space-valve numbering to capture local spatiotemporal coupling features. A time-frequency attention gating mechanism is an adaptive weight allocation unit that simultaneously focuses on frequency components and temporal dynamics to highlight key perturbation patterns. A bidirectional long short-term memory network is a recurrent network composed of forward and backward LSTMs, used to integrate past and future information to predict pressure evolution trends.

[0086] The physical channel calculation unit is an analytical calculation module that directly maps the difference between the target pressure value and the current pressure value to the basic valve opening based on fluid dynamics equations. The expression is as follows:

[0087] ,

[0088] in, This refers to the theoretically calculated value of the valve opening. It refers to the flow gain coefficient, which describes the linearity of the valve's flow curve; It refers to the hyperbolic tangent function; It refers to the pressure sensitivity factor, which is the reciprocal of the rock mass permeability. For example, the rock mass permeability of dense granite is 0.1 mD. This refers to the difference between the target pressure value and the current pressure value; This refers to the crack compensation aperture; This refers to the rock mass structure coefficient, which is determined by laboratory rock core calibration (e.g., 0.12 for granite). This refers to fracture density, which is statistically analyzed using borehole acoustic imaging. It refers to the average fracture aperture (mm), measured using a core microscope.

[0089] The residual channel calculation unit refers to a neural network submodule that weights and fuses the dynamic compensation vector and the pressure difference value through learnable parameters, and outputs the dynamic compensation opening component, as shown in the following expression:

[0090]

[0091] in, This refers to the dynamic compensation opening degree output by the LSTM network; This refers to the computation function of the Long Short-Term Memory neural network; This refers to the dynamic compensation vector.

[0092] The residual fusion formula is as follows:

[0093]

[0094] in, For the Sigmoid function; This refers to the reference valve opening degree; This refers to the upper limit of the valve's effective stroke (default 95%). This refers to the lower limit of the valve's effective stroke (default 5%).

[0095] The Sigmoid activation layer refers to a differentiable nonlinear function layer that compresses the fusion result to the 0-1 range to represent the valve stroke percentage. The mechanical limit protection layer refers to a hard constraint module that further trims the Sigmoid output to the 5%-95% range to prevent the actuator from exceeding its limits.

[0096] Please refer to Figure 2 , Figure 2 This is a schematic diagram of the module structure of the pressure-opening coupling model provided in Embodiment 1 of the single-hole pressure regulation method for the underground water-sealed cavern water curtain test of this application. The model includes three main modules: an acoustic feature extraction module, a temporal dynamic module, and a physical constraint decoding module. First, in the acoustic feature extraction module, one-dimensional convolutional networks with kernel lengths of 3, 7, and 11 are sequentially instantiated, and then attention gating layers, dimensionality reduction processing layers, and regression output layers are cascaded to complete module encapsulation. Then, a phase compensator, a three-dimensional convolutional layer, a time-frequency attention gating mechanism, and a bidirectional long short-term memory network are sequentially connected to form a temporal dynamic module. This module is used to process historical pressure sequences and historical valve action sequences, extract spatiotemporal features, and generate dynamic compensation vectors. Then, the physical channel calculation unit, residual channel calculation unit, sigmoid activation layer, and mechanical limit protection layer are connected in series to obtain the physical constraint decoding module. This module is responsible for performing dual-channel residual fusion calculation of hydraulic conduction correction coefficients, dynamic compensation vectors, target pressure values, and current pressure values ​​to finally obtain the reference valve opening. Finally, the input and output ports of the three modules are interconnected and encapsulated as a whole, thus completing the construction of the pressure-opening coupling model.

[0097] As an example, the training steps of the pressure-opening coupling model include: conducting a stepped flow rate variation experiment within a preset valve opening range with a preset step size; during the stepped flow rate variation experiment, collecting water curtain orifice pressure data, rock mass acoustic impedance spectrum data, and valve opening data; constructing a joint loss function that includes an opening error term, a rate of change constraint term, and a pressure reconstruction term; and training the pressure-opening coupling model using a backpropagation algorithm based on the water curtain orifice pressure data, the rock mass acoustic impedance spectrum data, and the valve opening data until the value of the joint loss function is less than a preset loss value.

[0098] The preset valve opening range refers to the safe and operable range of valve stroke limited to 10% to 90% during the model training phase to avoid mechanical dead zones and damage at extreme positions. The preset step size refers to the adjustment range in which the valve opening increases or decreases incrementally by a fixed 2% increment during each step flow variation test. The step flow variation test refers to generating repeatable flow-pressure step responses by discretely and progressively changing the valve opening to cover the dynamic characteristics of the entire operating range. Water curtain orifice pressure data refers to the sequence of instantaneous water pressure values ​​inside the water curtain orifice recorded in real time by a high-frequency pressure sensor during the test. Rock mass acoustic impedance spectrum data refers to the broadband acoustic impedance-frequency curve of the rock mass synchronously acquired at each valve opening level. Valve opening data refers to the current percentage of actual valve stroke fed back by the electric actuator. The opening error term refers to the squared loss of the difference between the model-predicted valve opening and the measured opening, used to constrain output accuracy. The rate of change constraint term is a penalty term when the valve opening change rate exceeds a set upper limit, used to suppress drastic actions. The pressure reconstruction term refers to the squared loss of the difference between the model-reconstructed pressure and the measured pressure, used to ensure the fidelity of the pressure response. The joint loss function is a single scalar loss formed by the weighted sum of the opening error term, the rate of change constraint term, and the pressure reconstruction term, used to unify the optimization objective. The formula is as follows:

[0099]

[0100] in, This refers to the total loss value; This refers to the valve opening predicted by the model; This refers to the actual opening degree of the valve; This refers to the rate of change of the opening degree over time; This refers to the pressure on the model calculations; This refers to target pressure.

[0101] First, 41 steady-state operating points are generated within the 10%–90% valve stroke range with a 2% step size. At each point, the PLC continuously writes the target opening three times at 1 Hz. After the actuator closes the loop and reaches its position, the steady-state condition is maintained for 30 seconds, while simultaneously recording the water curtain orifice pressure sampled at 100 Hz, the rock mass acoustic impedance spectrum obtained by a 10 kHz frequency sweep, and the actual valve opening feedback, forming a complete data frame with timestamps. This ensures time alignment of pressure, acoustics, and opening, resulting in a paired training set that covers the entire range and has repeatable noise. Second, the data is divided into training, validation, and testing subsets in an 8:1:1 ratio. The PyTorch DataLoader feeds the data into the model in batches of 256 frames. The forward propagation first outputs γ through the acoustic module, then δ through the timing module, and finally the predicted opening and reconstructed pressure are obtained through the physical constraint module. These are then used together with the measured values ​​to calculate the joint loss. The AdamW optimizer has an initial learning rate of 3e-4, decaying to 1e-5 every 20 epochs, with gradient clipping to 1.0 to prevent explosion. Finally, the training loop runs on a single RTX 3080 until the validation set loss does not decrease for 10 consecutive epochs, at which point it terminates. The weights with the minimum validation loss are saved, and the model is evaluated using the test set. If the error is <0.02 MPa and the opening error is <1%, the model is solidified as a torchscript file and deployed to the edge control box to achieve real-time inference.

[0102] Step S30: Adjust the valve of the water curtain orifice according to the reference valve opening, and monitor the pressure change acceleration of the water curtain orifice during the adjustment process.

[0103] It should be noted that the pressure change acceleration refers to the first derivative of the rate of change of the measured pressure inside the water curtain orifice per unit time, which is used to quantify the severity of pressure fluctuations.

[0104] Understandably, firstly, the controller converts the reference valve opening into a 4-20 mA drive current and writes it into the electric needle valve servo amplifier, causing the valve to reach that opening at a constant speed within 5 seconds. Then, it reads the orifice pressure sensor at a 1 kHz sampling rate, calculates the pressure difference between adjacent points in real time, and then performs another differential to obtain the instantaneous acceleration value. If the absolute value of the acceleration exceeds 0.02 MPa·s... - ²Then immediately record the timestamp and cache the data of the most recent 1 second for later determination of whether to trigger the compensation action.

[0105] Step S40: Calculate the opening compensation increment and damping attenuation coefficient based on the pressure change acceleration.

[0106] It should be noted that the opening compensation increment refers to the instantaneous stroke correction that needs to be superimposed on the base valve opening, calculated in real time based on the acceleration of pressure changes, to offset pressure fluctuations. The damping attenuation coefficient is an exponential weighting factor used to suppress pressure oscillations; its value increases with increasing acceleration and determines the rate of attenuation of the compensation increment in subsequent control cycles. The calculation formula is as follows:

[0107]

[0108] The basic damping value is 0.3; the upper limit threshold damping value is 0.9; and the energy scaling factor is 1.5. This refers to the damping attenuation coefficient; This refers to the oscillation energy density.

[0109] As an example, the steps of calculating the opening compensation increment and damping attenuation coefficient based on the pressure change acceleration include: when the pressure change acceleration is greater than a preset pressure acceleration threshold, acquiring pressure time series data, and generating a pressure deviation time series signal based on the pressure time series data and the target pressure value; performing a Hilbert-Huang transform on the pressure deviation time series signal to obtain a hydraulic transient component and a mechanical response hysteresis component; calculating the hysteresis time constant of the mechanical response hysteresis component through impact response filtering; calculating the opening compensation increment based on the valve's maximum adjustment rate, the current pressure deviation, and the hysteresis time constant; calculating the oscillation energy density of the hydraulic transient component, and generating a damping attenuation coefficient based on the oscillation energy density.

[0110] The preset pressure acceleration threshold refers to the critical acceleration value used to trigger the compensation algorithm. When the measured value exceeds this value, the pressure fluctuation is considered abnormal. Pressure time-series data refers to the numerical sequence of continuous changes in water curtain orifice pressure over time, recorded at a fixed sampling frequency. The pressure deviation time-series signal refers to the difference sequence obtained by subtracting the pressure time-series data from the target pressure value point by point, reflecting the degree of deviation and its evolution. The Hilbert-Huang transform is a time-frequency processing method that performs empirical mode decomposition on nonlinear and non-stationary signals followed by Hilbert spectral analysis. The hydraulic transient component refers to the rapid oscillation part of the pressure deviation signal caused by fluid inertia and compressibility, expressed as follows:

[0111]

[0112] in, This refers to the transient components of hydraulic power; This refers to the intrinsic mode frequencies; It refers to the i-th eigenmode function; t is the current time point.

[0113] The mechanical response hysteresis component refers to the delayed response portion of the pressure deviation signal caused by the mechanical inertia of the valve-piping system, and its expression is as follows:

[0114]

[0115] in, It refers to the mechanical response hysteresis component.

[0116] Impulse response filtering is a digital filtering technique that uses the unit impulse response function to deconvolve the hysteresis component of the mechanical response to extract pure hysteresis information. The hysteresis time constant is the time scale required for the hysteresis component of the mechanical response to reach 63.2% of its final change; it characterizes the system delay and is calculated using the following formula:

[0117]

[0118] in, This refers to the lag time constant; This refers to the peak pressure within the time window; It refers to the mechanical response hysteresis component.

[0119] The maximum valve regulation rate refers to the maximum percentage change in stroke allowed by the actuator per unit time, representing the upper limit of its actuation capacity. Current pressure deviation refers to the instantaneous difference between the measured pressure and the target pressure value at a given moment. Opening compensation increment is the instantaneous correction amount of the valve stroke calculated based on the current pressure deviation, hysteresis time constant, and the maximum valve regulation rate. The calculation formula is as follows:

[0120]

[0121] in, This represents the valve's maximum regulating rate. This refers to the incremental compensation for the opening degree.

[0122] Oscillation energy density refers to the energy distribution intensity of hydraulic transient components per unit time or per unit frequency in the time-frequency domain. It is used to quantify the severity of oscillations, and the calculation formula is as follows:

[0123]

[0124] in, This refers to the transient components of hydraulic power; It is the time normalization factor.

[0125] First, the dynamic adjustment system for the water curtain orifice calculates the pressure change acceleration in real time during each control cycle. Once it detects that the acceleration exceeds 0.02 MPa·s... -The system continuously acquires new pressure time-series data at a frequency of 1 kHz based on a preset threshold, and simultaneously reads the target pressure value. The two values ​​are then subtracted point by point to generate a pressure deviation time-series signal. This is done so that high-precision analysis is only initiated during abnormal fluctuations, avoiding computational waste during normal operation. The system then sends this deviation signal to the Hilbert-Huang transform thread, first decomposing it into several IMFs using EMD (Empirical Mode Decomposition), and then separating the high-frequency and low-frequency energies using the Hilbert spectrum to obtain the hydraulic transient component and the mechanical response hysteresis component. Subsequently, impact response filtering is performed on the mechanical response hysteresis component, and deconvolution is used to extract the moment when its amplitude decays to 63.2% as the hysteresis time constant. This step is used to quantify the pure delay of the valve-pipeline. Finally, the system calculates the opening compensation increment based on the current pressure deviation, hysteresis time constant, and the valve's maximum adjustment rate of 8% / s. At the same time, it integrates the hydraulic transient component in the 0–50 Hz frequency band to obtain the oscillation energy density and linearly maps it to a damping attenuation coefficient of 0.3–1.0. Both are immediately written into the valve servo command of the next control cycle to achieve overshoot suppression and rapid stabilization.

[0126] Step S50: Generate the target valve opening based on the reference valve opening, the opening compensation increment, and the damping attenuation coefficient.

[0127] It should be noted that the target valve opening refers to the final execution opening command value obtained by superimposing the reference valve opening with the opening compensation increment after being weighted by the damping attenuation coefficient within the current control cycle. This value is used to drive the valve to accurately reach the required pressure setting.

[0128] As an example, the step of generating the target valve opening based on the reference valve opening, the opening compensation increment, and the damping attenuation coefficient includes: calculating the damping constraint compensation amount based on the opening compensation increment and the damping attenuation coefficient; calculating the reference valve opening based on the reference valve opening and the damping constraint compensation amount; and applying range constraints to the reference valve opening to obtain the target valve opening.

[0129] The damping constraint compensation amount refers to the executable correction amount that gradually decays over time, obtained by exponentially weighting the opening compensation increment with the damping attenuation coefficient. The reference valve opening is the intermediate calculated value obtained by adding the baseline valve opening to the damping constraint compensation amount.

[0130] First, the opening compensation increment is multiplied by the damping attenuation coefficient to obtain the damping constraint compensation amount that decays exponentially with time. Then, this is added to the reference valve opening to obtain the reference valve opening. Finally, if the reference value is greater than 95%, the target valve opening is locked at 95%; if it is less than 5%, it is locked at 5%; and if it is between the two, the value is directly output, thus completing the generation of the target valve opening.

[0131] Step S60: Adjust the valve of the water curtain orifice to the target valve opening so that the water curtain orifice reaches the target pressure value.

[0132] Understandably, firstly, the water curtain orifice dynamic adjustment system converts the target valve opening into a 4-20 mA current signal, which is then used by a servo amplifier to drive the electric needle valve to move precisely to the corresponding stroke within 2 seconds. Subsequently, the orifice pressure is read in a closed loop and compared with the set value. If the deviation still exceeds ±0.05 MPa, the current is finely adjusted at 50 ms intervals until the measured pressure falls into the target range, thus completing one dynamic adjustment.

[0133] This embodiment provides a single-hole pressure regulation method for a water curtain test in an underground water-sealed cavern. First, the acoustic impedance spectrum of the rock mass is acquired and intrinsic mode decomposition is performed to extract geological feature components. This step allows real-time perception of the impact of internal rock fissures and water content changes on water seal performance, providing a geological basis for subsequent precise regulation. Next, a pressure-opening coupling model is used to process the target pressure value, current pressure value, and geological feature components to obtain the baseline valve opening. This model integrates acoustic feature extraction, temporal dynamic analysis, and physical constraint decoding, comprehensively considering the initial adjustment requirements of multiple factors on the valve opening. Then, the water curtain orifice valve is adjusted according to the baseline valve opening, and the acceleration of pressure changes is monitored. This step captures the intensity of pressure fluctuations in real time, providing immediate feedback for subsequent compensation. The opening compensation increment and damping attenuation coefficient are calculated based on the pressure change acceleration. This method allows for dynamic adjustment of the valve opening while suppressing hydraulic oscillations, ensuring the stability of pressure regulation. Finally, the target valve opening is generated based on the baseline valve opening, the opening compensation increment, and the damping attenuation coefficient, and the valve is adjusted to this opening, enabling the water curtain orifice to reach the target pressure value. This series of steps enables high-precision adaptive adjustment of the pressure in a single water curtain hole during underground water-sealed cavern water curtain tests, significantly improving test efficiency and safety.

[0134] Based on the first embodiment of this application, in the second embodiment of this application, the content that is the same as or similar to that in Embodiment 1 above can be referred to the above description, and will not be repeated hereafter. Based on this, please refer to... Figure 3 , Figure 3 This is a flowchart illustrating the second embodiment of the single-hole pressure regulation method for the underground water-sealed cavern water curtain test of this application. The pressure-opening coupling model includes an acoustic feature extraction module, a temporal dynamic module, and a physical constraint decoding module. Step S20 of the single-hole pressure regulation method for the underground water-sealed cavern water curtain test includes steps S21 to S23:

[0135] Step S21: The geological feature components are subjected to multi-scale convolutional encoding by the acoustic feature extraction module to obtain hydraulic conduction correction coefficients.

[0136] It should be noted that the hydraulic conduction correction coefficient is a scalar correction factor output after the geological feature components are multi-scale convolutionally encoded by the acoustic feature extraction module. It is used to quantify the influence of rock mass fractures and pore structures on water flow conductivity, so as to adjust the pressure-flow relationship of the water curtain orifice and make the control more precise.

[0137] As an example, the acoustic feature extraction module includes a one-dimensional convolutional network, an attention gating layer, a dimensionality reduction layer, and a regression output layer. The one-dimensional convolutional network includes a first convolutional layer, a second convolutional layer, and a third convolutional layer. The kernel length of the second convolutional layer is greater than the kernel length of the first convolutional layer, and the kernel length of the third convolutional layer is greater than the kernel length of the second convolutional layer. The step of performing multi-scale convolutional encoding on the geological feature components through the acoustic feature extraction module to obtain hydraulic conduction correction coefficients includes: performing high-frequency feature extraction on the geological feature components through the first convolutional layer to obtain microfracture features. The geological feature components are processed by a second convolutional layer to extract mid-frequency features, resulting in an interface effect feature vector. A third convolutional layer is used to extract low-frequency features, resulting in a permeability tensor. An attention-gated layer is used to weight and fuse the microfracture feature vector, the interface effect feature vector, and the permeability tensor to obtain a fused feature tensor. A dimensionality reduction layer is used to compress the fused feature tensor to obtain a low-dimensional feature vector. Finally, a regression output layer is used to linearly transform the low-dimensional feature vector to obtain a hydraulic conduction correction coefficient.

[0138] A one-dimensional convolutional network (BCNN) is a neural network structure composed of multiple one-dimensional convolutional layers, used to extract local features from sequential data. The first convolutional layer, with a short kernel length (e.g., 3), extracts high-frequency features from geological feature components, primarily reflecting the distribution of microfractures. The second convolutional layer, with a medium kernel length (e.g., 7), extracts mid-frequency features from geological feature components, primarily reflecting the effects of rock interface. The third convolutional layer, with a long kernel length (e.g., 11), extracts low-frequency features from geological feature components, primarily reflecting the permeability characteristics of the rock mass. The microfracture feature vector is the high-frequency feature vector extracted by the first convolutional layer, containing information about the distribution of microfractures in the rock mass, with an acoustic wavelength range of 0.1-1 m. The interface effect feature vector is the mid-frequency feature vector extracted by the second convolutional layer, reflecting the characteristics of rock interface, with an acoustic wavelength range of 1-5 m. The permeability tensor is the low-frequency feature tensor extracted by the third convolutional layer, containing information about the permeability characteristics of the rock mass, with an acoustic wavelength range >5 m. Attention-gated layers are network layers that weight and fuse feature vectors of different frequencies using learnable weights to highlight important features. The fused feature tensor is the feature tensor obtained after weighted fusion by the attention-gated layer, combining high-frequency, mid-frequency, and low-frequency features. Dimensionality reduction layers are network layers that perform channel compression on the fused feature tensor to reduce feature dimensionality and computational complexity. Low-dimensional feature vectors are feature vectors after dimensionality reduction, retaining key information. Regression output layers are network layers that map low-dimensional feature vectors to hydraulic conduction correction coefficients, outputting the final result.

[0139] First, geological feature components are fed in parallel into one-dimensional convolutional layers with kernel lengths of 3, 7, and 11, outputting high-frequency microfracture feature vectors of length 128 and stride 1, mid-frequency interface effect feature vectors of length 128, and permeability characteristic tensors of size 128×3, respectively. This multi-scale parallel extraction captures geological information at different spatial frequencies. Second, the three features are concatenated along the channel dimension and input into an attention-gated layer. Each channel feature is scored using Sigmoid gating weights and then weighted and summed to generate a fused feature tensor of size 128, automatically amplifying key information and suppressing redundant information. Subsequently, global average pooling is used to compress the 128 channels into a 16-dimensional low-dimensional feature vector, reducing parameters while retaining discriminative power. Finally, the 16-dimensional vector is fed into the fully connected weight matrix W (1×16) and bias b of the regression output layer, and a linear transformation is performed to obtain scalar hydraulic conduction correction coefficients, which are directly used for real-time correction of the subsequent pressure-aperture model, simplifying calculations and ensuring accuracy.

[0140] Step S22: Extract spatiotemporal features from the historical pressure sequence and historical valve action sequence using the time-series dynamic module to obtain a dynamic compensation vector.

[0141] It should be noted that the historical pressure sequence refers to a continuous sequence of water curtain orifice pressure values ​​recorded at a fixed sampling frequency over a past period. The historical valve action sequence refers to a continuous sequence of values, time-aligned with the historical pressure sequence, recording the changes in valve opening over time. The dynamic compensation vector refers to a one-dimensional numerical vector extracted from the above two sequences by the time-series dynamic module, used for real-time correction of the reference valve opening.

[0142] As an example, the temporal dynamic module includes a phase compensator, a first bidirectional long short-term memory network, a three-dimensional convolutional layer, a time-frequency attention gating mechanism, and a second bidirectional long short-term memory network. The step of extracting spatiotemporal features from historical pressure sequences and historical valve action sequences using the temporal dynamic module to obtain a dynamic compensation vector includes: performing sensor delay correction on the historical pressure sequence using the phase compensator to obtain a phase-corrected pressure sequence; and extracting temporal features from the phase-corrected pressure sequence using the first bidirectional long short-term memory network to obtain a temporal pressure feature vector. The first bidirectional long short-term memory network is configured with a first preset time step and a first preset dimension. The system employs hidden units; spatial encoding of historical valve action sequences using the three-dimensional convolutional layer yields spatiotemporal features of valve action; weighted fusion of the temporal pressure feature vector and the spatiotemporal features of valve action is achieved using the time-frequency attention gating mechanism to obtain pressure-valve action fusion features; and decision feature extraction of the pressure-valve action fusion features is performed using the second bidirectional long short-term memory network to obtain a dynamic compensation vector. The second bidirectional long short-term memory network is configured with hidden units of a second preset time step and a second preset dimension, as well as an acoustic gating mechanism. The second preset time step is smaller than the first preset time step, and the second preset dimension is smaller than the first preset dimension.

[0143] A phase compensator is a digital filtering unit that performs time-aligned filtering on historical pressure sequences to compensate for sensor delays. The phase compensation formula is as follows:

[0144]

[0145] Among them, the pressure sensor has a fixed delay of 0.2 seconds. This refers to the pressure change value after compensation; This refers to the original pressure measurement value; This refers to the rate of change of pressure.

[0146] The first bidirectional Long Short-Term Memory (LSTM) network refers to a bidirectional LSTM with 72 time steps and 64 hidden units, used to extract temporal features of the pressure sequence after phase correction. The 3D convolutional layer refers to a network layer that slides convolutional kernels across the time-space-valve channel 3D tensor to encode the spatiotemporal features of valve action. The time-frequency attention gating mechanism is an attention module that weights and fuses temporal pressure features and valve action features according to frequency and time weights. The second bidirectional LSTM network refers to a bidirectional LSTM with 36 time steps, 32 hidden units, and acoustic gating, used to output a dynamic compensation vector. The phase-corrected pressure sequence refers to the continuous pressure numerical sequence after delay correction by the phase compensator. The temporal pressure feature vector refers to the 64-dimensional pressure temporal representation vector output by the first bidirectional LSTM network. The first preset time step refers to the length of the time window of 72 sampling points. The first preset dimension refers to the 64-dimensional vector space of the LSTM hidden state. The hidden unit refers to the number of neurons inside the LSTM used for memory and information transmission. The valve action spatiotemporal features refer to the valve action tensor output by the 3D convolutional layer, which contains coupled temporal and spatial features. The pressure-valve action fusion features refer to the comprehensive feature tensor after weighted fusion using a time-frequency attention gating mechanism. The dynamic compensation vector refers to the one-dimensional numerical vector output by the second bidirectional long short-term memory network, used to correct the valve opening. The second preset time step refers to the short-time window length of 36 sampling points. The second preset dimension refers to the 32-dimensional vector space of the LSTM hidden states. The acoustic gating mechanism refers to the gating unit that dynamically adjusts the weights of the LSTM forget gate and input gate based on acoustic feature components, expressed as follows:

[0147]

[0148] Where γ is the hydraulic transmission correction factor; This refers to the output of the gate control signal; This refers to the Sigmoid activation function; This refers to the gating weight matrix; This refers to the hidden state at the previous moment; This refers to the input at the current moment.

[0149] First, the dynamic adjustment system for the water curtain orifice feeds the cached 72-point historical pressure sequence into a phase compensator. A 128th-order FIR filter is used for point-by-point convolution to align the sensor delays to the current moment, outputting a 72-point phase correction sequence to eliminate closed-loop errors caused by time misalignment. Second, this sequence is fed in parallel into a first bidirectional LSTM: 64-dimensional hidden units in both the forward and backward directions, fully expanded with a 72-step stride. A gated loop is used to calculate and output a 64-dimensional temporal pressure feature vector. Simultaneously, the valve action sequence corresponding to the 72-step stride is reshaped into a (72, 4, 1) tensor. A 4×4×1 local block is extracted using a 16-channel 3×3×3 three-dimensional convolution kernel, resulting in a 4×4×16 spatiotemporal feature of the valve action. This simultaneously captures the temporal action pattern coupled with the spatial valve arrangement. Then, the 64-dimensional pressure vector and the 4×4×16 valve features are flattened and concatenated, and then input into a time-frequency attention-gated system: first, global average pooling is performed on the channels to obtain a 16-dimensional weight vector, which is then activated by a Sigmoid function and multiplied back by the original features to complete the weighted fusion and output a fused feature tensor, ensuring that key time-frequency regions are amplified. Finally, this fused feature is fed into a second bidirectional LSTM: a 36-step window is pruned, a 32-dimensional hidden unit is used, and an acoustic gating mechanism is introduced at each time step (the 32-dimensional acoustic feature components are used to generate gating weights via a Sigmoid function to adjust the forget gate and the input gate). A 1-dimensional dynamic compensation vector is output through a single-layer feedforward. The short-time, small-dimensional design reduces the computational load and enables a fast response to the latest perturbations.

[0150] Step S23: The physical constraint decoding module performs dual-channel residual fusion calculation on the hydraulic transmission correction coefficient, the dynamic compensation vector, the target pressure value of the water curtain test, and the current pressure value of the water curtain orifice to obtain the reference valve opening.

[0151] As an example, the physical constraint decoding module includes a physical channel calculation unit, a residual channel calculation unit, a Sigmoid activation layer, and a mechanical limit protection layer. The step of obtaining the reference valve opening by performing dual-channel residual fusion calculation on the hydraulic conduction correction coefficient, the dynamic compensation vector, the target pressure value of the water curtain test, and the current pressure value of the water curtain orifice through the physical constraint decoding module includes: performing a nonlinear transformation on the pressure difference between the target pressure value of the water curtain test and the current pressure value of the water curtain orifice through the physical channel calculation unit to obtain a basic opening component; mapping the dynamic compensation vector and the pressure difference through the residual channel calculation unit to obtain a dynamic compensation opening component; performing amplitude limiting processing on the dynamic compensation opening component through the Sigmoid activation layer to obtain a compressed compensation component; superimposing the basic opening component and the compressed compensation component, and constraining the superposition result through the mechanical limit protection layer to obtain the reference valve opening.

[0152] The basic opening component refers to the initial valve opening value calculated solely based on the pressure difference, output by the physical channel calculation unit. Dynamically compensated opening component. This refers to the compensation value output by the residual channel calculation unit, used to correct the base opening. (Compression compensation component) This refers to the dynamically compensated aperture component after being limited by the Sigmoid activation layer. The calculation formula is as follows:

[0153]

[0154] First, the water curtain orifice dynamic adjustment system subtracts the current pressure value from the target pressure value to obtain ΔP. This scalar is then fed into the physical channel calculation unit. Using a pre-stored quadratic polynomial u_base = a·ΔP² + b·ΔP + c (coefficients a, b, and c are determined offline by calibration data), the basic opening component is output in the FPGA using single-precision floating-point operations, ensuring that the steady-state mapping conforms to the laws of fluid mechanics. Second, the same ΔP is concatenated with a 32-dimensional dynamic compensation vector to form a 33-dimensional input, which is then fed into the fully connected layer of the residual channel calculation unit (weight matrix 64×33, bias 64-dimensional). After ReLU activation, it is fully connected again and mapped to 1 dimension to obtain the dynamic compensation opening component, which is used to correct hysteresis and disturbances that the physical model cannot cover in real time. Finally, a Sigmoid activation layer is used to compress the dynamic compensation opening component into the 0–1 range to prevent excessive compensation amplitude from causing valve vibration. Finally, the basic opening component and the compression compensation component are added together, and the result is sent to the mechanical limit protection layer: if the value is >95%, it is forced to be 95%; if it is <5%, it is forced to be 5%; otherwise, it is directly output to obtain the reference valve opening and immediately write it into the valve servo drive to ensure that the command is both fast and safe.

[0155] This embodiment first uses an acoustic feature extraction module to perform multi-scale convolutional encoding on geological feature components to obtain hydraulic conduction correction coefficients. This step can reflect the influence of rock mass fissures on water flow in real time, providing a geological basis for subsequent regulation. Next, a time-series dynamic module extracts spatiotemporal features from historical pressure sequences and historical valve action sequences to obtain dynamic compensation vectors, which are used to capture system hysteresis and disturbances. Finally, a physical constraint decoding module performs dual-channel residual fusion calculation on the hydraulic conduction correction coefficients, dynamic compensation vectors, target pressure values, and current pressure values ​​to obtain the baseline valve opening. This ensures that the valve opening conforms to physical laws and adapts to dynamic changes, thereby achieving high-precision adaptive regulation of the water curtain orifice pressure.

[0156] For example, to help understand the implementation process of the single-hole pressure regulation method for the underground water-sealed cavern water curtain test obtained in this embodiment combined with the above-described embodiment one, please refer to... Figure 4 , Figure 4A time-series flowchart of a single-hole pressure regulation method for a water curtain test in an underground water-sealed cavern is provided, specifically:

[0157] This diagram illustrates the timing flow of a pressure-opening control system, comprising four parts: sensors, control unit, pressure-opening model, and valve actuator. In the initialization phase, the sensors acquire rock mass acoustic impedance spectrum data and return it to the control unit. The control unit extracts geological feature components through intrinsic mode decomposition. Next, in the baseline opening calculation phase, the control unit inputs the target pressure value / current pressure value / geological feature components. After calculation of γ by the acoustic module, δ by the timing module, and dual-channel fusion by the physical module, the baseline valve opening K_base is obtained and returned to the control unit. In the dynamic adjustment phase, the control unit executes the K_base opening command, while the sensors monitor the pressure change acceleration dP / dt² in real time and return the result. If the acceleration exceeds a threshold, the control unit calculates the compensation increment ΔK_c and the damping coefficient ζ, generating the target opening K_final = K_base + ζΔK_c. Finally, the control unit executes the K_final opening. If the acceleration is normal, the K_base opening is maintained. The entire process achieves high-precision adaptive adjustment of the water curtain orifice pressure through valve action feedback closed-loop control.

[0158] This application also provides a single-hole pressure regulating device for underground water-sealed cavern water curtain tests. Please refer to... Figure 5 The single-hole pressure regulating device for the underground water-sealed cavern water curtain test includes:

[0159] The mode decomposition module 10 is used to acquire the acoustic impedance spectrum of the rock mass and perform intrinsic mode decomposition on the acoustic impedance spectrum of the rock mass to extract geological feature components.

[0160] The model processing module 20 is used to process the target pressure value of the water curtain test, the current pressure value of the water curtain orifice, and the geological feature components through the pressure-opening coupling model to obtain the reference valve opening. The pressure-opening coupling model is constructed based on the acoustic feature extraction module, the temporal dynamic module, and the physical constraint decoding module.

[0161] The first valve adjustment module 30 is used to adjust the valve of the water curtain hole according to the reference valve opening, and to monitor the pressure change acceleration of the water curtain hole during the adjustment process;

[0162] Compensation increment calculation module 40 is used to calculate the opening compensation increment and damping attenuation coefficient based on the pressure change acceleration;

[0163] The valve opening calculation module 50 is used to generate a target valve opening based on the reference valve opening, the opening compensation increment, and the damping attenuation coefficient.

[0164] The second valve adjustment module 60 is used to adjust the valve of the water curtain orifice to the target valve opening so that the water curtain orifice reaches the target pressure value.

[0165] The single-hole pressure regulating device for underground water-sealed cavern water curtain tests provided in this application adopts the single-hole pressure regulating method for underground water-sealed cavern water curtain tests described in the above embodiments, and can solve the technical problem of how to achieve high-precision adaptive regulation of the pressure of a single water curtain hole in underground water-sealed cavern water curtain tests. Compared with the prior art, the beneficial effects of the single-hole pressure regulating device for underground water-sealed cavern water curtain tests provided in this application are the same as the beneficial effects of the single-hole pressure regulating method for underground water-sealed cavern water curtain tests provided in the above embodiments, and other technical features in the single-hole pressure regulating device for underground water-sealed cavern water curtain tests are the same as the features disclosed in the methods of the above embodiments, and will not be repeated here.

[0166] This application provides a single-hole pressure regulating device for a water curtain test in an underground water-sealed cavern. The single-hole pressure regulating device for a water curtain test in an underground water-sealed cavern includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the single-hole pressure regulating method for a water curtain test in an underground water-sealed cavern as described in Embodiment 1 above.

[0167] The following is for reference. Figure 6 The diagram illustrates a structural schematic of a single-hole pressure regulating device suitable for implementing a water curtain test in an underground water-sealed cavern according to embodiments of this application. The single-hole pressure regulating device for the water curtain test in an underground water-sealed cavern according to embodiments of this application may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital radio receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Description), PMPs (Portable Media Players), and vehicle terminals (e.g., vehicle navigation terminals), as well as fixed terminals such as digital TVs and desktop computers. Figure 6 The single-hole pressure regulating device shown in the underground water-sealed cavern water curtain test is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0168] like Figure 6As shown, the single-hole pressure regulating device for the underground water-sealed cavern water curtain test may include a processing unit 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to the program stored in ROM (Read Only Memory) 1002 or the program loaded from storage device 1003 into RAM (Random Access Memory) 1004. RAM 1004 also stores various programs and data required for the operation of the single-hole pressure regulating device for the underground water-sealed cavern water curtain test. The processing unit 1001, ROM 1002, and RAM 1004 are interconnected via bus 1005. Input / output (I / O) interface 1006 is also connected to the bus. Typically, the following systems can be connected to I / O interface 1006: input devices 1007 including, for example, touchscreens, touchpads, keyboards, mice, image sensors, microphones, accelerometers, gyroscopes, etc.; output devices 1008 including, for example, LCDs (Liquid Crystal Displays), speakers, vibrators, etc.; storage devices 1003 including, for example, magnetic tapes, hard disks, etc.; and communication devices 1009. Communication device 1009 allows the single-hole pressure regulating device for underground water-sealed cavern water curtain tests to exchange data wirelessly or via wired communication with other devices. Although the figure shows a single-hole pressure regulating device for underground water-sealed cavern water curtain tests with various systems, it should be understood that it is not required to implement or possess all the systems shown. More or fewer systems may be implemented alternatively.

[0169] Specifically, according to the embodiments disclosed in this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments disclosed in this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from storage device 1003, or installed from ROM 1002. When the computer program is executed by processing device 1001, it performs the functions defined in the methods of the embodiments disclosed in this application.

[0170] The single-hole pressure regulating device for underground water-sealed cavern water curtain tests provided in this application adopts the single-hole pressure regulating method for underground water-sealed cavern water curtain tests described in the above embodiments, and can solve the technical problem of how to achieve high-precision adaptive regulation of the pressure of a single water curtain hole in underground water-sealed cavern water curtain tests. Compared with the prior art, the beneficial effects of the single-hole pressure regulating device for underground water-sealed cavern water curtain tests provided in this application are the same as the beneficial effects of the single-hole pressure regulating method for underground water-sealed cavern water curtain tests provided in the above embodiments, and other technical features of the single-hole pressure regulating device for underground water-sealed cavern water curtain tests are the same as those disclosed in the previous embodiment method, and will not be repeated here.

[0171] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0172] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0173] This application provides a computer-readable storage medium having computer-readable program instructions (i.e., a computer program) stored thereon, the computer-readable program instructions being used to execute the single-hole pressure regulation method for the underground water-sealed cavern water curtain test in the above embodiments.

[0174] The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, RAM (Random Access Memory), ROM (Read Only Memory), EPROM (Erasable Programmable Read Only Memory or Flash Memory), optical fibers, CD-ROM (CD-Read Only Memory), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, system, or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination thereof.

[0175] The aforementioned computer-readable storage medium may be included in the single-hole pressure regulating device of the underground water-sealed cavern water curtain test; or it may exist independently and not be assembled into the single-hole pressure regulating device of the underground water-sealed cavern water curtain test. The aforementioned computer-readable storage medium carries one or more programs. When these programs are executed by the single-hole pressure regulating device for the underground water-sealed cavern water curtain test, the single-hole pressure regulating device performs the following: acquires the rock mass acoustic impedance spectrum and performs intrinsic mode decomposition on the rock mass acoustic impedance spectrum to extract geological feature components; processes the target pressure value of the water curtain test, the current pressure value of the water curtain orifice, and the geological feature components through a pressure-opening coupling model to obtain a reference valve opening, wherein the pressure-opening coupling model is constructed based on an acoustic feature extraction module, a temporal dynamic module, and a physical constraint decoding module; adjusts the valve of the water curtain orifice according to the reference valve opening and monitors the pressure change acceleration of the water curtain orifice during the adjustment process; calculates the opening compensation increment and damping attenuation coefficient based on the pressure change acceleration; generates the target valve opening based on the reference valve opening, the opening compensation increment, and the damping attenuation coefficient; and adjusts the valve of the water curtain orifice to the target valve opening so that the water curtain orifice reaches the target pressure value.

[0176] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including LAN (Local Area Network) or WAN (Wide Area Network)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0177] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0178] The readable storage medium provided in this application is a computer-readable storage medium that stores computer-readable program instructions (i.e., a computer program) for executing the single-hole pressure regulation method of the above-described underground water-sealed cavern water curtain test. This solves the technical problem of how to achieve high-precision adaptive regulation of the pressure of a single water curtain hole in an underground water-sealed cavern water curtain test. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as those of the single-hole pressure regulation method of the underground water-sealed cavern water curtain test provided in the above embodiments, and will not be elaborated upon here.

[0179] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the single-hole pressure regulation method for underground water-sealed cavern water curtain tests as described above. The computer program product provided by this application solves the technical problem of how to achieve high-precision adaptive regulation of the pressure of a single water curtain orifice in underground water-sealed cavern water curtain tests. Compared with the prior art, the beneficial effects of the computer program product provided by this application are the same as those of the single-hole pressure regulation method for underground water-sealed cavern water curtain tests provided in the above embodiments, and will not be repeated here.

[0180] The above description is only a part of the embodiments of this application and does not limit the patent scope of this application. All equivalent structural transformations made under the technical concept of this application and using the contents of the specification and drawings of this application, or direct / indirect applications in other related technical fields, are included in the patent protection scope of this application.

Claims

1. A method for regulating single-hole pressure in an underground water-sealed cavern water curtain test, characterized in that, The method includes: Acoustic impedance spectra of the rock mass were collected, and intrinsic mode decomposition was performed on the rock mass acoustic impedance spectra to extract geological feature components. The target pressure value, the current pressure value of the water curtain orifice, and the geological feature components of the water curtain test are processed by a pressure-opening coupling model to obtain the reference valve opening. The pressure-opening coupling model is constructed based on the acoustic feature extraction module, the temporal dynamic module, and the physical constraint decoding module. The valve of the water curtain orifice is adjusted according to the reference valve opening, and the pressure change acceleration of the water curtain orifice is monitored during the adjustment process; Calculate the opening compensation increment and damping attenuation coefficient based on the pressure change acceleration; The target valve opening is generated based on the reference valve opening, the opening compensation increment, and the damping attenuation coefficient. Adjust the valve of the water curtain orifice to the target valve opening so that the water curtain orifice reaches the target pressure value; The step of processing the target pressure value of the water curtain test, the current pressure value of the water curtain orifice, and the geological feature components through a pressure-opening coupling model to obtain the reference valve opening includes: The geological feature components are multi-scale convolutional encoded by the acoustic feature extraction module to obtain hydraulic conduction correction coefficients. The time-series dynamic module extracts spatiotemporal features from historical pressure sequences and historical valve action sequences to obtain dynamic compensation vectors. The physical constraint decoding module performs dual-channel residual fusion calculation on the hydraulic transmission correction coefficient, the dynamic compensation vector, the target pressure value of the water curtain test, and the current pressure value of the water curtain orifice to obtain the reference valve opening.

2. The method as described in claim 1, characterized in that, The steps for calculating the opening compensation increment and damping attenuation coefficient based on the pressure change acceleration include: When the pressure change acceleration is greater than a preset pressure acceleration threshold, pressure time series data is acquired, and a pressure deviation time series signal is generated based on the pressure time series data and the target pressure value. The pressure deviation time-series signal is subjected to Hilbert-Huang transform to obtain the hydraulic transient component and the mechanical response hysteresis component; The lag time constant of the mechanical response lag component is calculated by shock response filtering; The valve opening compensation increment is calculated based on the valve's maximum regulating rate, the current pressure deviation, and the hysteresis time constant. Calculate the oscillation energy density of the hydraulic transient component, and generate a damping attenuation coefficient based on the oscillation energy density.

3. The method as described in claim 1, characterized in that, The steps for constructing the pressure-opening coupling model include: An acoustic feature extraction module is constructed based on a one-dimensional convolutional network, an attention gating layer, a dimensionality reduction processing layer, and a regression output layer. A temporal dynamic module is constructed based on a phase compensator, a three-dimensional convolutional layer, a time-frequency attention gating mechanism, and a bidirectional long short-term memory network. A physical constraint decoding module is constructed based on the physical channel calculation unit, the residual channel calculation unit, the Sigmoid activation layer, and the mechanical limit protection layer. By combining the acoustic feature extraction module, the temporal dynamics module, and the physical constraint decoding module, a pressure-opening coupling model is obtained.

4. The method as described in claim 1, characterized in that, The acoustic feature extraction module includes a one-dimensional convolutional network, an attention gating layer, a dimensionality reduction processing layer, and a regression output layer. The one-dimensional convolutional network includes a first convolutional layer, a second convolutional layer, and a third convolutional layer. The kernel length of the second convolutional layer is greater than the kernel length of the first convolutional layer, and the kernel length of the third convolutional layer is greater than the kernel length of the second convolutional layer. The step of performing multi-scale convolutional encoding on the geological feature components through the acoustic feature extraction module to obtain hydraulic conduction correction coefficients includes: The geological feature components are subjected to high-frequency feature extraction through the first convolutional layer to obtain the microfracture feature vector; The geological feature components are extracted using the second convolutional layer to obtain the interface effect feature vector. The permeability tensor is obtained by extracting low-frequency features from the geological feature components through the third convolutional layer. The attention gating layer is used to weight and fuse the microcrack feature vector, the interface effect feature vector, and the permeability tensor to obtain a fused feature tensor. The fusion feature tensor is channel-compressed through the dimensionality reduction layer to obtain a low-dimensional feature vector; The hydraulic conduction correction coefficient is obtained by performing a linear transformation on the low-dimensional feature vector through the regression output layer.

5. The method as described in claim 1, characterized in that, The time-series dynamic module includes a phase compensator, a first bidirectional long short-term memory network, a three-dimensional convolutional layer, a time-frequency attention gating mechanism, and a second bidirectional long short-term memory network. The step of extracting spatiotemporal features from historical pressure sequences and historical valve action sequences using the time-series dynamic module to obtain dynamic compensation vectors includes: The phase compensator is used to perform sensor delay correction on the historical pressure sequence to obtain the phase-corrected pressure sequence. The phase-corrected stress sequence is subjected to temporal feature extraction through the first bidirectional long short-term memory network to obtain a temporal stress feature vector. The first bidirectional long short-term memory network is configured with hidden units of a first preset time step and a first preset dimension. The historical valve action sequence is spatially encoded by the three-dimensional convolutional layer to obtain the spatiotemporal features of valve action; The time-frequency attention gating mechanism is used to weight and fuse the temporal pressure feature vector and the spatiotemporal features of valve action to obtain pressure-valve action fusion features. The second bidirectional long short-term memory network is used to extract decision features from the pressure-valve action fusion features to obtain a dynamic compensation vector. The second bidirectional long short-term memory network is configured with a second preset time step, a second preset dimension of hidden units, and an acoustic gating mechanism. The second preset time step is smaller than the first preset time step, and the second preset dimension is smaller than the first preset dimension.

6. The method as described in claim 1, characterized in that, The physical constraint decoding module includes a physical channel calculation unit, a residual channel calculation unit, a Sigmoid activation layer, and a mechanical limit protection layer. The step of obtaining the reference valve opening by performing dual-channel residual fusion calculation on the hydraulic conduction correction coefficient, the dynamic compensation vector, the target pressure value of the water curtain test, and the current pressure value of the water curtain orifice through the physical constraint decoding module includes: The physical channel calculation unit performs a nonlinear transformation on the pressure difference between the target pressure value of the water curtain test and the current pressure value of the water curtain orifice to obtain the basic opening component. The dynamic compensation opening component is obtained by mapping the dynamic compensation vector and the pressure difference value through the residual channel calculation unit. The dynamic compensation aperture component is subjected to amplitude limiting processing through the Sigmoid activation layer to obtain the compressed compensation component; The reference valve opening is obtained by superimposing the basic opening component and the compression compensation component, and constraining the superposition result through the mechanical limit protection layer.

7. A single-hole pressure regulating device for underground water-sealed cavern water curtain tests, characterized in that, The device includes: The mode decomposition module is used to acquire the acoustic impedance spectrum of the rock mass and perform intrinsic mode decomposition on the acoustic impedance spectrum of the rock mass to extract geological feature components. The model processing module is used to process the target pressure value of the water curtain test, the current pressure value of the water curtain orifice, and the geological feature components through a pressure-opening coupling model to obtain the reference valve opening. The pressure-opening coupling model is constructed based on the acoustic feature extraction module, the temporal dynamics module, and the physical constraint decoding module. The steps of processing the target pressure value of the water curtain test, the current pressure value of the water curtain orifice, and the geological feature components through the pressure-opening coupling model to obtain the reference valve opening include: performing multi-scale convolutional encoding on the geological feature components through the acoustic feature extraction module to obtain hydraulic conduction correction coefficients; performing spatiotemporal feature extraction on historical pressure sequences and historical valve action sequences through the temporal dynamics module to obtain dynamic compensation vectors; and performing dual-channel residual fusion calculation on the hydraulic conduction correction coefficients, the dynamic compensation vectors, the target pressure value of the water curtain test, and the current pressure value of the water curtain orifice through the physical constraint decoding module to obtain the reference valve opening. The first valve adjustment module is used to adjust the valve of the water curtain orifice according to the reference valve opening, and to monitor the pressure change acceleration of the water curtain orifice during the adjustment process; The compensation increment calculation module is used to calculate the opening compensation increment and damping attenuation coefficient based on the pressure change acceleration. The valve opening calculation module is used to generate the target valve opening based on the reference valve opening, the opening compensation increment, and the damping attenuation coefficient. The second valve adjustment module is used to adjust the valve of the water curtain orifice to the target valve opening so that the water curtain orifice reaches the target pressure value.

8. A single-hole pressure regulating device for underground water-sealed cavern water curtain tests, characterized in that, The device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the single-hole pressure regulation method for the underground water-sealed cavern water curtain test as described in any one of claims 1 to 6.

9. A storage medium, characterized in that, The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, it implements the steps of the single-hole pressure regulation method for the underground water-sealed cavern water curtain test as described in any one of claims 1 to 6.

Images