An intelligent grouting control method and system suitable for high-altitude complex geology

Abstract

The application provides a kind of intelligent grouting control method and system suitable for high altitude complex geology, it is related to automatic control technical field, comprising: S1, the characteristic data of current grouting section is obtained, the characteristic data is mapped to the pre-judgment geological type by pre-trained machine learning model, and the control parameter is initialized according to the pre-judgment geological type;S2, from initial grouting pressure, gradually increase pressure according to reference step;S3, the characteristic data is monitored by running monitoring window;S4, the attenuation slope reference is dynamically established based on the attenuation slope of previous several monitoring windows, and the current attenuation slope is compared with the attenuation slope reference, and the response state is determined in combination with sensitivity factor;S5, according to the determination result of response state, the coupling adjustment of control parameter is executed, to actively verify the geological type;S6, according to the geological type confirmed to the characteristic data of current grouting section is added label, the pre-judgment accuracy of geological type and the adaptability of control parameter are continuously improved.

Description

Technical Field

[0001] This invention relates to the field of automatic control technology, specifically to an intelligent grouting control method and system suitable for complex geological conditions at high altitudes. Background Technology

[0002] During grouting construction, the complex geological conditions in high-altitude areas make the dynamic characteristics of the controlled object (rock fissures) highly uncertain, which places extremely high demands on the adaptive adjustment capability of the grouting control system. Although existing grouting control methods have achieved automation to a certain extent, their core control strategies still use fixed observation window lengths and fixed pressure increment steps, that is, they lack a mechanism to dynamically adjust the control parameters (sampling period and control increment) according to the real-time response of the controlled object. Such fixed parameter control modes have the following common problems: (1) When the fissure shows a rapid decrease in grout absorption (the controlled object responds quickly), the fixed long observation window will cause unnecessary control waiting and reduce adjustment efficiency; while when the fissure shows a slow decrease in grout absorption (the controlled object responds slowly), the fixed short observation window cannot accurately identify the decay trend due to insufficient observation time, resulting in a lag in control decision; (2) During the control process, the rapid decrease in injection rate may reflect the true state of small fissure filling or may originate from the temporary closure of large fissures under pressure disturbance (i.e., external disturbance signal). The existing control system lacks an active verification mechanism and cannot effectively identify the true state of the controlled object and external disturbances, which can easily lead to misjudgment; (3) The fixed long observation window is slow to respond to transient anomalies such as splitting or uplift that are very likely to occur in high-altitude strata, and cannot generate protection control commands (such as pressure reduction) in a timely manner, which can easily cause system instability or even engineering accidents. Summary of the Invention

[0003] The purpose of this invention is to provide an intelligent grouting control method and system suitable for complex geological conditions at high altitudes. The geological conditions of the grouting area are regarded as the controlled object of the control system, and the grouting pressure and injection rate are regarded as the input and feedback signals of the control system. By monitoring the response status of the controlled object in real time, the control parameters are adaptively adjusted to achieve dynamic identification and precise control of the controlled object.

[0004] To solve the above-mentioned technical problems, the present invention adopts the following solution: A smart grouting control method suitable for high-altitude complex geological conditions includes: S1. During the water pressure stage, acquire the feature data of the current grouting section, map the feature data to the predicted geological type through a pre-trained machine learning model, and initialize the control parameters according to the predicted geological type; the control parameters include the initial grouting pressure, the reference step size, and the sensitivity factor. S2. Starting from the initial grouting pressure, gradually increase the pressure according to the reference step length to grout the current grouting section; S3. During the grouting stage, the operation monitoring window monitors the characteristic data of the current grouting section; S4. Based on the attenuation slope of the previous monitoring windows, a dynamic attenuation slope benchmark is established, and the current attenuation slope is compared with the attenuation slope benchmark. Combined with the sensitivity factor, the response state is determined. S5. Based on the result of the response status determination, perform coupled adjustment of the control parameters: If the response is determined to be in a rapid decay state, the baseline step size for the next step is reduced and the monitoring window for the next step is extended. If the response is still in a rapid decay state within the monitoring window after the coupling adjustment, the geological type is confirmed to be a small fracture. If the response is determined to be in a slow decay state, the baseline step size for the next step is increased and the monitoring window for the next step is shortened. If the response is still in a slow decay state within the monitoring window after the coupling adjustment, the geological type is confirmed to be a large fissure or a karst cave. S6. Add labels to the feature data of the current grouting section based on the confirmed geological type for incremental training or parameter optimization of the subsequent machine learning model.

[0005] Furthermore, the feature data includes: the relationship curve between pressure and injection rate obtained during the water pressure stage; the fluctuation characteristics of the relationship curve between pressure and injection rate are identified by a pre-trained machine learning model to determine whether the geological type is dense, small-fracture, or complex, and the sensitivity factor is increased sequentially according to the order of dense, small-fracture, and complex to relax the judgment boundary of the response state.

[0006] Furthermore, in S4, the process of determining the response status is as follows: Calculate the current decay slope of the injection rate in the current monitoring window, and call the previous few monitoring windows of the current monitoring window to calculate the decay slope of the injection rate in the previous few monitoring windows respectively. In this way, the mean and standard deviation of the decay slope are calculated. The difference between the mean and the sensitivity factor multiplied by the standard deviation is used as the boundary for determining the rapid decay state, and the sum of the mean and the sensitivity factor multiplied by the standard deviation is used as the boundary for determining the slow decay state. When the current attenuation slope is less than the threshold for determining the rapid attenuation state, it is determined to be in a rapid attenuation state; when the current attenuation slope is greater than the threshold for determining the slow attenuation state, it is determined to be in a slow attenuation state.

[0007] Furthermore, in S5, when a rapid decay state is determined, the reference step size for the next step is reduced by 30% to 70% of the original step size, and the monitoring window for the next step is extended by 1.2 to 2 times the original duration; when a slow decay state is determined, the reference step size for the next step is increased by 1.2 to 1.5 times the original step size, and the monitoring window for the next step is shortened by 50% to 70% of the original duration.

[0008] Furthermore, in S5, if a rapid decay state is determined, within the coupled and adjusted monitoring window, if the response state is not a rapid decay state and the current injection rate is higher than the preset threshold, it is determined to be a disturbance signal; if a slow decay state is determined, within the coupled and adjusted monitoring window, if the response state is not a slow decay state and the current injection rate is lower than the preset threshold, it is determined that the small crack filling is close to saturation, and the reference step size is restored to continue increasing the pressure.

[0009] Furthermore, in S3, the monitoring window in the running monitoring window is a long-term window; while running the long-term window, a short-term window can also be run in parallel to monitor the characteristic data of the current grouting section. The sampling frequency of the short time window is higher than that of the long time window, and it is used to capture transient signals of pressure or injection rate. During the monitoring process, if the short time window detects a transient signal, it executes a pressure reduction protection command.

[0010] Furthermore, transient signals identified within a short time window include: if the pressure drops below a first threshold and the injection rate rises above a second threshold within a unit of time, the transient signal is determined to be formation fracturing; or, if the pressure rises above a third threshold and the injection rate drops to zero within a unit of time, the transient signal is determined to be pipeline blockage.

[0011] Furthermore, the machine learning model employs one of random forest, XGBoost, or lightweight neural network, and uses the geological type labels obtained from actual verification during the grouting process as the target variable for incremental training.

[0012] A smart grouting control system suitable for high-altitude complex geology, employing a smart grouting control method suitable for high-altitude complex geology, comprising: Initialization module: During the water pressure stage, the feature data of the current grouting section is acquired, and the feature data is mapped to the predicted geological type through a pre-trained machine learning model. The control parameters are initialized according to the predicted geological type. The control parameters include the initial grouting pressure, the baseline step size, and the sensitivity factor. Control quantity application module: Starting from the initial grouting pressure, gradually increase the pressure according to the reference step size to grout the current grouting section; Monitoring module: During the grouting stage, the monitoring window monitors the characteristic data of the current grouting section; Status determination module: Based on the attenuation slope of the previous monitoring windows, a dynamic attenuation slope benchmark is established, and the current attenuation slope is compared with the attenuation slope benchmark. Combined with the sensitivity factor, the response status is determined. Active verification module: Based on the response status determination, performs coupled adjustments to control parameters. If the response is determined to be in a rapid decay state, the baseline step size for the next step is reduced and the monitoring window for the next step is extended. If the response is still in a rapid decay state within the monitoring window after the coupling adjustment, the geological type is confirmed to be a small fracture. If the response is determined to be in a slow decay state, the baseline step size for the next step is increased and the monitoring window for the next step is shortened. If the response is still in a slow decay state within the monitoring window after the coupling adjustment, the geological type is confirmed to be a large fissure or a karst cave. Feedback module: Labels are added to the feature data of the current grouting section based on the confirmed geological type, which are used for incremental training or parameter optimization of subsequent machine learning models.

[0013] The beneficial effects of this invention are: This invention provides an intelligent grouting control method suitable for complex geological conditions at high altitudes. Compared with existing technologies that rely on experience to set fixed parameters, this invention proposes to input water pressure test data into a pre-trained machine learning model. Before grouting begins, the geological type (dense, small-fracture, or complex) is predicted, and the initial grouting pressure, baseline step size, and sensitivity factor are initialized accordingly. This gives the system prior predictive capabilities, reducing ineffective trial time in the early stages of grouting and significantly improving construction efficiency. Furthermore, the actively verified geological type is added as a label to the water pressure feature data of the current grouting section for incremental training or parameter optimization of the subsequent machine learning model. As engineering data accumulates, the system's accuracy in predicting geological types and the adaptability of control parameters continuously improve, forming a data-driven closed-loop optimization.

[0014] Meanwhile, this invention, through a hierarchical control architecture that combines feedforward prediction and feedback verification, enables the control system to automatically tune and continuously optimize control parameters under conditions where the dynamic characteristics of the controlled object are unknown, significantly improving the control system's adaptability to complex and uncertain environments. Attached Figure Description

[0015] Figure 1 This is a flowchart illustrating the intelligent grouting control method in Embodiment 1 of the present invention.

[0016] Figure 2 This is a first schematic diagram of the coupling adjustment after performing the fast decay branch in Embodiment 1 of the present invention.

[0017] Figure 3This is a second schematic diagram showing the coupling adjustment after performing the fast decay branch in Embodiment 1 of the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The following description of at least one exemplary embodiment is merely illustrative and is in no way intended to limit the present invention or its application or use. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] Unless otherwise specifically stated, the relative arrangement, numerical expressions, and values ​​of the components and steps described in these embodiments do not limit the scope of the invention.

[0020] At the same time, it should be understood that, for ease of description, the dimensions of the various parts shown in the accompanying drawings are not drawn according to actual scale.

[0021] Furthermore, for clarity and brevity, descriptions of well-known structures, functions, and configurations may have been omitted. Those skilled in the art will recognize that various changes and modifications can be made to the examples described herein without departing from the spirit and scope of this disclosure.

[0022] Techniques, methods, and equipment known to those skilled in the art may not be discussed in detail, but where appropriate, such techniques, methods, and equipment should be considered part of the specification.

[0023] In all examples shown and discussed herein, any specific values ​​should be interpreted as merely exemplary and not as limitations. Therefore, other examples of exemplary embodiments may have different values.

[0024] The present invention will now be described in detail with reference to the accompanying drawings and embodiments: Example 1 In this embodiment, a smart grouting control method suitable for high-altitude and complex geological conditions is provided. This smart grouting control method is as follows: Figure 1 As shown, it includes the following steps: S1. During the water pressure stage, acquire the feature data of the current grouting section, map the feature data to the predicted geological type through a pre-trained machine learning model, and initialize the control parameters according to the predicted geological type; the control parameters include the initial grouting pressure, the reference step size, and the sensitivity factor. S2. Starting from the initial grouting pressure, gradually increase the pressure according to the reference step length to grout the current grouting section; S3. During the grouting stage, the operation monitoring window monitors the characteristic data of the current grouting section; S4. Based on the attenuation slope of the previous monitoring windows, a dynamic attenuation slope benchmark is established, and the current attenuation slope is compared with the attenuation slope benchmark. Combined with the sensitivity factor, the response state is determined. S5. Based on the response status determination, perform coupling adjustments to actively verify the geological type: If the response is determined to be in a rapid decay state, the baseline step size for the next step is reduced and the monitoring window for the next step is extended. If the response is still in a rapid decay state within the monitoring window after the coupling adjustment, the geological type is confirmed to be a small fracture. If the response is determined to be in a slow decay state, the baseline step size for the next step is increased and the monitoring window for the next step is shortened. If the response is still in a slow decay state within the monitoring window after the coupling adjustment, the geological type is confirmed to be a large fissure or a karst cave. S6. Add labels to the feature data of the current grouting section based on the confirmed geological type for incremental training or parameter optimization of the subsequent machine learning model.

[0025] Based on the above principles, the present invention will be further described as follows: First, it should be noted that this embodiment employs a strategy of coordinating and adjusting the monitoring window duration with the baseline step size, rather than independently adjusting a single parameter. The principle is that pressure changes in the grouting system (determined by the step size) and flow response observations (determined by the window size) are strongly correlated. When the step size decreases (pressure rise slows), the magnitude of the injection rate change itself weakens. If the window is not extended at this time, the signal change captured within the short window will be lower than the sensor resolution, causing the effective signal to be submerged by noise. Conversely, when the step size increases, pressure disturbances intensify. If the window is too long, minor formation fracturing symptoms will be smoothed out by the averaging effect within a long window. Therefore, this invention treats the step size and window size as a pair of control parameters and adjusts them in the same direction: decreasing the step size extends the window in the same direction, ensuring sufficient observation time to accumulate effective statistics under low excitation intensity; increasing the step size shortens the window in the same direction, ensuring rapid capture of transient responses under high excitation intensity. This coupling strategy is a reasonable engineering design made by those skilled in the art based on the technical inspiration of this invention, considering the characteristics of the controlled object in grouting (high inertia, strong noise), rather than variable confusion.

[0026] This embodiment uses the grouting project of a high-altitude hydropower station as an example for illustration. The dam site of this project is at an altitude of 3200 meters, with complex geological conditions. The rock mass is mainly limestone, with uneven development of fissures, including small fissures, wide fissures, and karst caves. The design grouting pressure for the second grouting section of borehole No. 3 is 2.5 MPa, and the completion criterion is that the injection rate is ≤1 L / min under the design pressure and lasts for 30 minutes.

[0027] First, step S1 is executed: During the water pressure test, characteristic data of the current grouting section is acquired. This characteristic data includes the permeability value obtained during the water pressure test and the relationship curve between pressure and injection rate. Specifically, the permeability value measured in this section's water pressure test is 4.2 Lu, and the relationship curve between pressure and injection rate shows a low-slope, gently rising type. A pre-trained machine learning model (using the XGBoost algorithm) maps the characteristic data to a predicted geological type. The model is trained using 2000 sets of water pressure characteristic data from historical grouting projects and corresponding geological type sample sets. The model outputs a predicted geological type of small fractures.

[0028] Here, a three-category classification (dense, small-fracture, and complex) is used in the preliminary assessment stage. The aim is to roughly estimate the permeability characteristics of the rock mass by observing the macroscopic morphology of the pressure curve before grouting begins, thus setting a reasonable initial sensitivity factor for subsequent control. These three categories physically represent a permeability gradient from low to high: dense indicates extremely low permeability, small-fracture represents moderate fracture development, and complex indicates extremely uneven fracture distribution or the presence of large channels. Once grouting has begun, the control system no longer requires such fine differentiation; the core indicator that truly determines the control strategy is the rate of decay of the injection rate. Both dense and small-fracture types exhibit a rapid decrease in injection rate to zero under pressure (the former due to lack of grouting sites, and the latter due to rapid filling of small spaces), thus both can be identified as small-fracture types during the verification stage. Complex types, due to their chaotic fracture distribution and the presence of large-scale channels, have an ineffective injection rate establishment, exhibiting a slow decay, and are therefore identified as large-fracture or cavern types during the verification stage.

[0029] Specifically, control parameters are initialized based on the predicted geological type. These control parameters include initial grouting pressure, baseline step size, and sensitivity factor. The initial grouting pressure is set to 20% of the design pressure, i.e., 0.5 MPa; the baseline step size is 0.15 MPa; and the sensitivity factor is 1.5.

[0030] Next, proceed to step S2: starting from the initial grouting pressure of 0.5MPa, gradually increase the pressure according to the reference step size of 0.15MPa to grout the current grouting section.

[0031] Step S3: During the grouting stage, the monitoring window monitors the characteristic data of the current grouting section. In this embodiment, the monitoring window includes a long-term window and a parallel short-term window. The long-term window is set to 30 seconds in length and has a sampling frequency of 1 Hz, used to periodically calculate the decay slope of the injection rate; the short-term window has a sampling frequency of 20 Hz, used to capture transient signals of pressure or injection rate, and executes a pressure reduction protection command when a transient signal is detected.

[0032] Step S4: Dynamically establish an attenuation slope benchmark based on the attenuation slopes of the previous monitoring windows. In the initial stage of grouting, the system collects injection rate attenuation slope data for the first five long-term windows, which are -3.2, -3.5, -3.3, -3.6, and -3.4, respectively. The control system calculates the mean attenuation slope as -3.4 and the standard deviation as 0.15. Combining this with a sensitivity factor of 1.5, the difference between the mean and the sensitivity factor multiplied by the standard deviation is used as the boundary for determining the rapid attenuation state, and the sum of the mean and the sensitivity factor multiplied by the standard deviation is used as the boundary for determining the slow attenuation state. The calculated rapid attenuation boundary is -3.625, and the slow attenuation boundary is -3.175. At the end of the sixth long-term window, the current attenuation slope is -3.8. Since the current attenuation slope is less than the boundary for determining the rapid attenuation state, the control system determines the current response state as a rapid attenuation state.

[0033] Step S5: Based on the response status determination, perform coupling adjustments to actively verify the geological type. Since the system is determined to be in a rapid decay state, it performs coupling adjustments on the rapid decay branch: reducing the baseline step size for the next step and extending the monitoring window for the next step. Specifically, as follows... Figure 2 As shown, the baseline step size for the next step was reduced from 0.15 MPa to 0.05 MPa (a reduction of approximately 66.7%), and the time window for the next step was extended from 30 s to 50 s (an extension of approximately 1.67 times). Within the coupled and adjusted monitoring window, the system continuously observed the injection rate. Observational data showed that the initial injection rate was 22 L / min, and the final injection rate was 9 L / min, still showing a continuous rapid decay trend. After recalculating the decay slope, the response state remained in a rapid decay state, thus confirming the geological type as small fractures. If, within the coupled and adjusted monitoring window, the response state changes from rapid decay to non-rapid decay and the current injection rate is higher than the preset threshold, it is determined to be a false signal of pressure disturbance.

[0034] Then, step S6 is executed: Based on the confirmed geological type of small fractures, labels are added to the characteristic data of the current grouting section (permeability value of 4.2 Lu, pressure-injection rate curve showing a gradual increase) for incremental training or parameter optimization of the subsequent machine learning model. The system restores the baseline step size of 0.15 MPa and the conventional long time window of 30 s, repeating steps S2 to S6, gradually increasing the pressure to the corrected upper limit of the design pressure of 2.125 MPa, and stabilizing the pressure at the design pressure until the injection rate is ≤1 L / min for 30 min, at which point grouting ends. Here, considering the lower atmospheric pressure in high-altitude areas, the upper limit of the design pressure is corrected according to engineering specifications to prevent formation fracturing, with a correction factor of 0.85.

[0035] In one embodiment, the same steps S1 to S6 are performed on the grouting section. The water pressure characteristic data shows a permeability value of 18.6, and the relationship curve between pressure and injection rate exhibits a steep upward slope. The machine learning model output predicts a complex geological type. Initialized control parameters: initial grouting pressure 0.3 MPa, baseline step size 0.10 MPa, sensitivity factor 2.5 (range 1.0~3.0). High altitude correction also applies, with the upper limit of design pressure adjusted to 2.125 MPa. The attenuation slopes for the first five long-term windows in the initial stage of grouting are: -1.1, -1.3, -1.0, -1.2, -1.1. The calculated mean is -1.14, and the standard deviation is 0.11. The rapid attenuation judgment boundary is -1.415, and the slow attenuation judgment boundary is -0.865. The current window attenuation slope is -0.7, which is greater than the slow attenuation judgment boundary, and is judged to be in a slow attenuation state. The coupling adjustment of the slow decay branch was performed: the baseline step size for the next step was increased and the monitoring window for the next step was shortened. The baseline step size was increased from 0.10 MPa to 0.13 MPa (an increase of 30%), and the long window was shortened from 30 s to 15 s (a reduction of 50%). Within the adjusted window, the injection rate decreased only from 35 L / min to 32 L / min, showing no significant decay. The decay slope was recalculated and remained in a slow decay state, confirming the geological type as a large fracture or karst cave. The area was then labeled as a large fracture / karst cave for incremental training. The pressure was gradually increased to the upper limit of the design pressure, and the termination condition was met.

[0036] If, after coupling adjustment, the response state changes from slow decay to non-slow decay and the injection rate is lower than a preset threshold, it is determined that the small crack filling is close to saturation. The system can then trigger a flow-limiting strategy to control the injection rate below 15 L / min and prepare for thickening of the slurry.

[0037] In one embodiment, the grouting section is monitored and initially determined to be in a rapid decay state (current decay slope -3.9), such as... Figure 3 As shown, the coupling adjustment of the rapid decay branch is performed: the baseline step size is reduced from 0.15 MPa to 0.05 MPa, and the long time window is extended from 30 s to 50 s. Within the adjusted 50 s window, the injection rate decreases from 20 L / min to 18 L / min and then gradually recovers to 22 L / min, the decay slope becomes -0.5 (non-rapid decay), and the current injection rate of 22 L / min is higher than the preset threshold of 5 L / min (the preset threshold is 5 L / min to 10 L / min). At this point, it is determined to be a disturbance signal, and the current grouting pressure is locked (the pressure is kept constant, and further pressure increases are prohibited). The system issues an early warning, prompting the operator to check the formation condition.

[0038] In one embodiment, during the normal pressurization process of the grouting section, a short-time window detected a sudden pressure increase of 0.6 MPa and a sudden drop in injection rate from 18 L / min to 4 L / min, lasting approximately 0.2 seconds, detected at a sampling frequency of 20 Hz. Based on the function of the short-time window, the control system immediately determined it to be a pipeline blockage event and executed a pressure reduction protection command to prevent high pressure from damaging the equipment. On-site inspection revealed that the blockage was caused by a bend in the orifice pipeline; after the blockage was cleared, grouting resumed.

[0039] In summary, after continuous construction of multiple grouting sections, the system can accumulate new, validated samples composed of feature data and geological types. Lightweight incremental training is automatically triggered in the background, employing one of the following: Random Forest, XGBoost, or a lightweight neural network, using the actually validated geological type labels as the target variable for incremental training. The updated model significantly improves the prediction accuracy for subsequent grouting sections.

[0040] Example 2 A smart grouting control system suitable for high-altitude complex geology, employing a smart grouting control method suitable for high-altitude complex geology, comprising: Initialization module: During the water pressure stage, the feature data of the current grouting section is acquired, and the feature data is mapped to the predicted geological type through a pre-trained machine learning model. The control parameters are initialized according to the predicted geological type. The control parameters include the initial grouting pressure, the baseline step size, and the sensitivity factor. Control quantity application module: Starting from the initial grouting pressure, gradually increase the pressure according to the reference step size to grout the current grouting section; Monitoring module: During the grouting stage, the monitoring window monitors the characteristic data of the current grouting section; Status determination module: Based on the attenuation slope of the previous monitoring windows, a dynamic attenuation slope benchmark is established, and the current attenuation slope is compared with the attenuation slope benchmark. Combined with the sensitivity factor, the response status is determined. Active verification module: Based on the response status determination, it performs coupling adjustments to actively verify the geological type. If the response is determined to be in a rapid decay state, the baseline step size for the next step is reduced and the monitoring window for the next step is extended. If the response is still in a rapid decay state within the monitoring window after the coupling adjustment, the geological type is confirmed to be a small fracture. If the response is determined to be in a slow decay state, the baseline step size for the next step is increased and the monitoring window for the next step is shortened. If the response is still in a slow decay state within the monitoring window after the coupling adjustment, the geological type is confirmed to be a large fissure or a karst cave. Feedback module: Labels are added to the feature data of the current grouting section based on the confirmed geological type, which are used for incremental training or parameter optimization of subsequent machine learning models.

[0041] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Based on the technical essence of the present invention, any simple modifications, equivalent substitutions, and improvements made to the above embodiments within the spirit and principles of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A smart grouting control method suitable for high-altitude complex geological conditions, characterized in that, include: S1. During the water pressure stage, acquire the feature data of the current grouting section, map the feature data to the predicted geological type through a pre-trained machine learning model, and initialize the control parameters according to the predicted geological type; the control parameters include the initial grouting pressure, the reference step size, and the sensitivity factor. S2. Starting from the initial grouting pressure, gradually increase the pressure according to the reference step length to grout the current grouting section; S3. During the grouting stage, activate the monitoring window to monitor the characteristic data of the current grouting section; S4. Based on the attenuation slope of the previous monitoring windows, a dynamic attenuation slope benchmark is established, and the current attenuation slope is compared with the attenuation slope benchmark. Combined with the sensitivity factor, the response state is determined. S5. Based on the result of the response status determination, perform coupled adjustment of the control parameters: If the state is determined to be a rapid decay state, the first control command is generated to reduce the reference step size for the next step and extend the monitoring window for the next step. If the response state is still a rapid decay state within the coupled and adjusted monitoring window, the geological type is confirmed to be a small fracture. If the response is determined to be in a slow decay state, a second control command is generated to increase the baseline step size for the next step and shorten the monitoring window for the next step. If the response is still in a slow decay state within the coupled and adjusted monitoring window, the geological type is confirmed to be a large fissure or a karst cave. S6. Add labels to the feature data of the current grouting section based on the confirmed geological type for incremental training or parameter optimization of the subsequent machine learning model.

2. The intelligent grouting control method applicable to complex geological conditions at high altitudes according to claim 1, characterized in that, The feature data includes: the relationship curve between pressure and injection rate obtained during the water pressure stage; the fluctuation characteristics of the relationship curve between pressure and injection rate are identified by a pre-trained machine learning model to determine whether the geological type is dense, small-fracture, or complex, and the sensitivity factor is increased sequentially according to the order of dense, small-fracture, and complex to relax the judgment boundary of the response state.

3. The intelligent grouting control method applicable to complex geological conditions at high altitudes according to claim 2, characterized in that, In S4, the process of determining the response status is as follows: Calculate the current decay slope of the injection rate in the current monitoring window, and call the previous few monitoring windows of the current monitoring window to calculate the decay slope of the injection rate in the previous few monitoring windows respectively. In this way, the mean and standard deviation of the decay slope are calculated. The difference between the mean and the sensitivity factor multiplied by the standard deviation is used as the boundary for determining the rapid decay state, and the sum of the mean and the sensitivity factor multiplied by the standard deviation is used as the boundary for determining the slow decay state. When the current attenuation slope is less than the threshold for determining a rapid attenuation state, it is determined to be a rapid attenuation state; when the current attenuation slope is greater than the threshold for determining a slow attenuation state, it is determined to be a slow attenuation state.

4. The intelligent grouting control method applicable to complex geological conditions at high altitudes according to claim 1, characterized in that, In S5, when a rapid decay state is determined, the reference step size of the next step is reduced by 30% to 70% of the original step size, and the monitoring window of the next step is extended by 1.2 to 2 times the original duration. When the state is determined to be slow decay, the ratio of increasing the reference step size of the next step is 1.2 to 1.5 times the original step size, and the ratio of shortening the monitoring window of the next step is 50% to 70% of the original duration.

5. The intelligent grouting control method applicable to complex geological conditions at high altitudes according to claim 1, characterized in that, In S5, if the state is determined to be fast decay, within the monitoring window after coupling adjustment, if the response state is not fast decay and the current injection rate is higher than the preset threshold, it is determined to be a disturbance signal; if the state is determined to be slow decay, within the monitoring window after coupling adjustment, if the response state is not slow decay and the current injection rate is lower than the preset threshold, it is determined to be small crack filling saturation, and the base step size is restored to continue increasing the pressure.

6. The intelligent grouting control method applicable to complex geological conditions at high altitudes according to claim 1, characterized in that, In S3, the monitoring window in the running monitoring window is a long-term window; while running the long-term window, a short-term window can also be run in parallel to monitor the characteristic data of the current grouting section; The sampling frequency of the short time window is higher than that of the long time window, and it is used to capture transient signals of pressure or injection rate. During the monitoring process, if the short time window detects a transient signal, it executes a pressure reduction protection command.

7. The intelligent grouting control method applicable to complex geological conditions at high altitudes according to claim 6, characterized in that, Transient signals identified within a short time window include: if the pressure drops by more than a first threshold and the injection rate rises by more than a second threshold within a unit time, the transient signal is determined to be formation fracturing; or, if the pressure rises by more than a third threshold and the injection rate drops to zero within a unit time, the transient signal is determined to be pipeline blockage.

8. The intelligent grouting control method applicable to complex geological conditions at high altitudes according to claim 1, characterized in that, The machine learning model employs one of random forest, XGBoost, or lightweight neural network, and uses the geological type labels obtained from actual verification during the grouting process as the target variable for incremental training.

9. An intelligent grouting control system suitable for high-altitude and complex geological conditions, characterized in that, The application of the intelligent grouting control method for high-altitude complex geology as described in any one of claims 1-8 includes: Initialization module: During the water pressure stage, the feature data of the current grouting section is acquired, and the feature data is mapped to the predicted geological type through a pre-trained machine learning model. The control parameters are initialized according to the predicted geological type. The control parameters include the initial grouting pressure, the baseline step size, and the sensitivity factor. Control quantity application module: Starting from the initial grouting pressure, gradually increase the pressure according to the reference step size to grout the current grouting section; Monitoring module: During the grouting stage, the monitoring window monitors the characteristic data of the current grouting section; Status determination module: Based on the attenuation slope of the previous monitoring windows, a dynamic attenuation slope benchmark is established, and the current attenuation slope is compared with the attenuation slope benchmark. Combined with the sensitivity factor, the response status is determined. Active verification module: Based on the response status determination, performs coupled adjustments to control parameters. If the state is determined to be a rapid decay state, the first control command is generated to reduce the reference step size for the next step and extend the monitoring window for the next step. If the response state is still a rapid decay state within the coupled and adjusted monitoring window, the geological type is confirmed to be a small fracture. If the response is determined to be in a slow decay state, a second control command is generated to increase the baseline step size for the next step and shorten the monitoring window for the next step. If the response is still in a slow decay state within the coupled and adjusted monitoring window, the geological type is confirmed to be a large fissure or a karst cave. Feedback module: Labels are added to the feature data of the current grouting section based on the confirmed geological type, which are used for incremental training or parameter optimization of subsequent machine learning models.

Images