Offshore wind turbine pile foundation cured soil injection system and method

Abstract

The present application relates to a kind of offshore wind turbine pile foundation solidified soil perfusion system and method, belong to offshore wind power construction technical field, including underwater sensing unit, pumping and execution unit and perfusion control unit, underwater sensing unit is composed of unmanned remote control submersible and its carried multi-beam imaging sonar, profile sonar, mud height gauge and flow velocity and direction sensor, realize the comprehensive perception of underwater environment, pumping and execution unit includes end ±90 ° swing and by submersible remote control electric control rotating conduit, perfusion control unit generates deviation distribution diagram by comparing real-time mud elevation data with three-dimensional prefill model point by point, and based on PID or model predictive control algorithm, the residence time, swing direction and lifting speed of conduit are automatically adjusted, submersible is connected to the front of conduit by releasable gripper and positioned accurately guided.The present application realizes the full-process closed-loop intelligent control of scour pit repair, significantly improves perfusion quality, efficiency and safety.

Description

Technical Field

[0001] This invention relates to the field of offshore wind power construction technology, and in particular to an offshore wind turbine pile foundation solidification soil grouting system and method. Background Technology

[0002] In complex hydrodynamic environments, the seabed around the piles of offshore wind turbine monopile or jacket foundations is prone to forming scour pits several meters or even more than ten meters deep due to ocean currents. Scour pits significantly weaken the lateral bearing capacity and vertical support stiffness of the pile foundation, seriously threatening the safe operation of the wind turbine. Solidified soil (such as underwater anti-dispersion cement-based materials) grouting is currently the mainstream scour pit repair technology.

[0003] The existing solidified soil grouting construction method has the following technical defects:

[0004] 1. High degree of blindness: Traditional construction relies on divers or simple ROV video guidance, which cannot accurately perceive the underwater topography and the real-time pattern of grout diffusion in turbid water. The injection point and volume depend on experience, which can easily lead to incomplete filling or overfilling.

[0005] 2. Lack of adaptive control: The duct cannot be dynamically adjusted according to the actual accumulation slope of the slurry underwater and the influence of ocean currents, resulting in deviation of the filling profile from the design and creating a risk of secondary scouring.

[0006] 3. Weak anomaly handling capability: There is a lack of automatic identification and intelligent handling mechanisms for emergencies such as pipe blockage, slurry dispersion by ocean currents, and ROV displacement in deep water environments, which often leads to construction interruptions or quality accidents.

[0007] 4. Post-assessment and refilling are disconnected: After the refilling is completed, there is a lack of a rapid and accurate mechanism for volume measurement and refilling decision-making, which requires the redeployment of measurement and operation vessels, resulting in low efficiency.

[0008] Therefore, there is an urgent need for a solidified soil grouting system and method that can achieve accurate perception, intelligent decision-making and adaptive control throughout the entire process. Summary of the Invention

[0009] This invention provides a system and method for grouting solidified soil for offshore wind turbine pile foundations, which solves the problems of poor sensing ability, low control precision, lack of adaptability and intelligence in the existing technology for grouting solidified soil in offshore wind turbine pile foundation scour pits.

[0010] The solution to the above-mentioned technical problems of the present invention is as follows: a soil grouting system and method for offshore wind turbine pile foundation, comprising an underwater sensing unit, a pumping and execution unit, and a grouting control unit. The underwater sensing unit includes an unmanned remotely operated vehicle (UAV) and a multibeam sonar, a profile sonar, and a mud level height meter mounted on the UAV, as well as a flow velocity and direction sensor installed near the duct opening. The multibeam sonar is used to construct underwater topography and slurry diffusion cloud maps in real time, the profile sonar is used to accurately measure the accumulation slope, the mud level height meter is used for distance measurement in turbid environments, and the flow velocity and direction sensor is used to sense the influence of ocean currents on the slurry, thereby achieving comprehensive and accurate perception of the underwater grouting environment.

[0011] The pumping and execution unit includes a high-pressure concrete pump, an electrically controlled rotary conduit, and a pressure sensor and a mud density meter installed inside the conduit. The end of the electrically controlled rotary conduit can swing ±90°, and its orientation can be remotely controlled by an unmanned remotely operated vehicle. The pressure sensor monitors the pumping pressure, the mud density meter monitors the uniformity of the slurry, and the electrically controlled rotary conduit enables precise adjustment of the injection direction.

[0012] The grouting control unit includes an industrial computer, a communication module, a memory storing a three-dimensional scour pit pre-filling model, a real-time data fusion and deviation analysis module, and an automatic decision-making algorithm module. The real-time data fusion and deviation analysis module is used to compare the real-time mud surface elevation data acquired by the underwater sensing unit with the three-dimensional scour pit pre-filling model point by point to generate a three-dimensional deviation distribution map. By fusing the pre-filling model with real-time data, it can accurately identify underfilling or overfilling areas during the grouting process, providing a quantitative basis for adaptive control.

[0013] The automatic decision-making algorithm module, based on the three-dimensional deviation distribution map and the flow velocity and direction data collected by the flow velocity and direction sensors, uses PID control or model predictive control to generate and output control commands for the pumping and execution units, so as to adjust the residence time, swing direction and lifting speed of the electrically controlled rotating conduit. This module realizes the autonomous optimization of the infusion strategy and can dynamically adjust the conduit residence time, swing direction, lifting speed and pumping parameters.

[0014] The unmanned remotely operated vehicle (UAV) is connected to the electrically controlled rotating duct via a detachable clamp, and the UAV is positioned in front of the duct for guidance, so as to achieve precise positioning and orientation of the duct tip. This structure ensures the spatial position accuracy of the duct tip in complex underwater environments and can be quickly released in emergency situations.

[0015] Furthermore, the control instructions generated by the automatic decision-making algorithm module for the pumping and execution unit include instructions to the high-pressure concrete pump to perform reverse pumping operation to handle pipe blockage events. This design achieves rapid and autonomous handling of pipe blockage events through an automatic reverse unblocking mechanism, effectively reducing manual intervention and construction interruption time, and lowering the risk of pipe bursts.

[0016] Furthermore, the irrigation control unit is also configured to automatically switch to using data from a mud level gauge to control the lifting height of the electrically controlled rotating duct when the multibeam imaging sonar or profiling sonar fails due to water turbidity. This feature utilizes sensor redundancy design to solve the problem of sonar sensing failure in turbid water, ensuring that the irrigation process can still be carried out continuously and reliably under harsh water quality conditions, and significantly enhancing the environmental adaptability of the system.

[0017] The present invention provides an intelligent grouting method for solidifying soil in offshore wind turbine pile foundations using the above-mentioned system, comprising the following contents.

[0018] Before construction, an unmanned surface vessel equipped with a multibeam echo sounder was used to conduct a full-coverage survey of the scour pits around the pile foundation, generating a high-precision digital terrain model. Based on the designed filling volume, the grouting blocks were divided in the digital twin platform, and a grouting path was planned starting from the deepest point and extending outward in a spiral. This step can accurately obtain the three-dimensional shape of the scour pits, providing accurate design benchmarks and construction paths for subsequent grouting, and avoiding blind grouting.

[0019] Subsequently, the unmanned remotely operated vehicle (UAV) and the electrically controlled rotating conduit are connected via a detachable clamp and submerged in water. The UAV acquires underwater topography and video in real time, and the end of the conduit automatically adjusts to the initial height from the bottom of the pit. This step achieves safe lowering and initial positioning of the system, ensuring the accuracy of the injection start position.

[0020] Layered grouting is performed. During each layer of grouting, multibeam imaging sonar continuously scans to acquire the actual diffusion pattern of the grout. The digital twin platform compares the actual filling elevation with the expected design elevation and automatically executes one or more of the following adaptive controls based on the comparison deviation: adjusting the residence time or oscillation speed of the guide tube in a specific area; instructing the guide tube to oscillate in the opposite direction of the ocean current to compensate for grout offset; when the accumulation height reaches the set value, prompting to move to the next grouting point. This step realizes closed-loop adaptive control of the grouting process, which can dynamically adjust the grouting strategy according to the real-time accumulation pattern, effectively improving the filling uniformity and contour accuracy.

[0021] During the grouting process, abnormal situations are monitored and handled automatically in real time. These abnormal situations include pipe blockage events caused by a sudden increase in pump pressure and a decrease in flow rate, excessive dispersion of grout, and loss of position of the unmanned remotely operated vehicle. This step significantly enhances the system's ability to respond autonomously to sudden working conditions, ensuring the continuity and safety of construction.

[0022] After the injection is completed, a 360° scan is performed using an unmanned remotely operated vehicle and multibeam imaging sonar to generate a post-fill terrain model. The volume of the unfilled area is automatically calculated. If the error exceeds the threshold, a decision is made and re-injection is performed. This step achieves seamless integration between the post-injection quality assessment and the re-injection operation, avoiding secondary measurements and equipment redeployment, and greatly improving operational efficiency.

[0023] Furthermore, during the layered grouting process, the lifting speed of the electrically controlled rotating guide pipe is determined in real time by a regression prediction model based on a BP neural network. Its input layer nodes correspond to the pump flow rate, slurry consistency, and current accumulation slope, while the output layer nodes correspond to the guide pipe lifting speed. The constraint target during model training is to control the accumulation slope between 1:10 and 1:15. This feature enables fine-grained control of the guide pipe lifting speed through an artificial intelligence model, ensuring that the slurry accumulation slope is always maintained within the designed optimal stable range. This not only guarantees the filling density but also prevents slope instability, significantly improving the stability and reliability of the grouting quality.

[0024] Furthermore, during the stratified grouting process, when the multibeam imaging sonar fails due to grout turbidity, it automatically switches to the data from the mud level gauge to control the lifting height of the guide tube. This feature utilizes sensor redundancy to effectively solve the problem of sonar sensing failure caused by water turbidity due to grouting disturbance, ensuring that grouting operations can still be carried out continuously and reliably under adverse visual conditions, and improving the system's adaptability to complex environments.

[0025] Furthermore, the automatic handling of the pipe blockage event is as follows: when a sudden increase in pump pressure exceeding 20% ​​and a simultaneous decrease in flow rate exceeding 15% are detected, the high-pressure concrete pump is instructed to perform a reverse pumping operation for 3-5 seconds; if the fault is not resolved after reversal, an automatic alarm is triggered and grouting is stopped. The automatic handling of the slurry over-dispersion event is as follows: when sonar detects that the underwater diffusion radius of the slurry is greater than twice the design value, the pumping pressure is automatically reduced, and the shipboard mixing system is instructed to increase the amount of admixtures to improve the slurry viscosity. The automatic handling of the unmanned remotely operated vehicle (UAV) dislocation event is as follows: when the UAV is detected to have detached from the conduit or lost its positioning signal, pumping is automatically suspended, and an acoustic beacon is activated to guide its retrieval. These quantitative and automated identification and handling logics comprehensively cover common sudden risks in grouting construction, significantly improve the robustness and operational safety of the system, and minimize unplanned downtime and material waste.

[0026] Furthermore, after the infilling is completed, if the calculated volume error of the unfilled area is greater than 3%, the system will automatically decide to refill; if the refilling volume is less than 5% of the total filling volume, the system will directly control the conduit to return to the target area for secondary infilling without replanning the global path. This feature implements a graded refilling strategy based on the amount of underfilling: for small-scale underfilling, rapid fixed-point refilling is used, while for large-scale underfilling, a complete refilling process is triggered. Under the premise of fully ensuring the quality of repair, the time and cost of refilling operations are reduced to the greatest extent, achieving a high balance between efficiency and quality.

[0027] Furthermore, the high-precision digital terrain model generated before construction has a resolution of 1cm, and the design filling volume includes an overfill coefficient of 1.05~1.10. The grouting blocks are divided into 3 to 5 layers according to the depth of the scour pit. High-resolution modeling ensures the accurate representation of the scour pit morphology. The overfill coefficient effectively compensates for the unavoidable material loss during underwater grouting, while layered grouting strictly controls the thickness of a single layer, avoiding grout slippage and poor interlayer bonding caused by excessively thick single layers, significantly improving the overall stability and compactness of the final filling body.

[0028] Furthermore, during the layered grouting process, the initial flow rate of each layer is calculated as the ratio of the designed volume to the expected grouting time. The scanning frequency of the multi-beam imaging sonar is one frame every 3 seconds. The reasonable initial flow rate setting avoids excessive grout diffusion caused by excessive flow rate, and also prevents construction efficiency from being affected by insufficient flow rate. The high-frequency sonar scanning provides timely and continuous data feedback for adaptive control, enabling the system to quickly respond to dynamic changes in the accumulation morphology, and significantly improving the real-time performance and accuracy of the control.

[0029] The beneficial effects of this invention are as follows: This invention provides a system and method for grouting solidified soil for offshore wind turbine pile foundations, which has the following advantages:

[0030] 1. By constructing an intelligent system that includes an underwater sensing unit, a pumping and execution unit, and an injection control unit, a closed-loop intelligent operation is realized, from precise surveying and modeling and path planning before construction, to real-time sensing, deviation analysis and adaptive control during injection, and automatic detection and graded re-injection after injection, which greatly reduces the reliance on human experience.

[0031] 2. The multi-sensor fusion and redundancy switching design ensures operational continuity under harsh sensing conditions such as turbid water, while the active directional compensation based on ocean current data effectively solves the problem of underwater slurry drift.

[0032] 3. By using a digital twin model for point-by-point elevation comparison and combining it with artificial intelligence model-driven control of the guide pipe lifting speed and dwell time, the precise shaping of the filling contour was achieved, avoiding underfilling or overfilling and ensuring the pile foundation repair effect.

[0033] 4. The quantitative automatic detection and handling logic designed for common accidents such as pipe blockage, slurry dispersion, and underwater robot misalignment significantly improves the robustness of the system and construction safety, and reduces unplanned downtime;

[0034] 5. Without replacing equipment, volume measurement can be performed immediately after grouting, and rapid replenishment can be carried out based on the level of underfill, which greatly improves the overall work efficiency and final repair quality.

[0035] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it according to the contents of the specification, the preferred embodiments of the present invention are described in detail below with reference to the accompanying drawings. Specific embodiments of the present invention are given in detail below with reference to the accompanying drawings. Attached Figure Description

[0036] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings:

[0037] Figure 1 This is a flowchart illustrating a method for a soil grouting system and method for solidifying offshore wind turbine pile foundations, provided in an embodiment of the present invention.

[0038] Figure 2 This is a system architecture diagram of a soil grouting system and method for offshore wind turbine pile foundations provided in an embodiment of the present invention. Detailed Implementation

[0039] The following is in conjunction with the appendix Figure 1-2 The principles and features of the present invention are described below. The examples given are for illustrative purposes only and are not intended to limit the scope of the invention. The invention is described more specifically in the following paragraphs by way of example with reference to the accompanying drawings. The advantages and features of the invention will become clearer from the following description. It should be noted that the drawings are in a very simplified form and use non-precise proportions, and are only used to facilitate and clarify the illustration of the embodiments of the invention.

[0040] It should be noted that when a component is described as "fixed to" another component, it can be directly on the other component or may have a component in between. When a component is considered "connected to" another component, it can be directly connected to the other component or may have a component in between. When a component is considered "set on" another component, it can be directly set on the other component or may have a component in between. The terms "vertical," "horizontal," "left," "right," and similar expressions used in this document are for illustrative purposes only.

[0041] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0042] Please see Figure 2 The present invention provides a soil grouting system for offshore wind turbine pile foundation, which includes an underwater sensing unit, a pumping and execution unit, and a grouting control unit.

[0043] Specifically, the underwater sensing unit includes an unmanned remotely operated vehicle (ROV) and a multibeam imaging sonar, a profile sonar, and a mud level altimeter mounted on the ROV, as well as a velocity and direction sensor installed near the duct opening. The multibeam imaging sonar is used to construct two-dimensional or three-dimensional cloud maps of underwater topography and slurry diffusion in real time; the profile sonar is used to accurately measure the contour of the solidified soil accumulation slope; the mud level altimeter is used to provide stable distance measurements from the bottom when the sonar signal is attenuated due to slurry turbidity; and the velocity and direction sensor is used to sense the influence of ocean currents on the slurry diffusion pattern in real time. Through the organic combination of the above multi-source sensors, a comprehensive and accurate perception of the underwater irrigation environment, from macroscopic topography to local flow field, is achieved.

[0044] The pumping and execution unit includes a high-pressure concrete pump located on a surface support platform (such as an engineering vessel), an electrically controlled rotary duct, and pressure sensors and a mud densitometer installed inside the duct. The end of the electrically controlled rotary duct can swing ±90°, and its swing direction and angle can be remotely controlled by the ROV through feedback from the underwater sensing unit. The pressure sensor is used to monitor the real-time pressure in the pumping circuit, and the mud densitometer is used to monitor the density of the pumped slurry, thereby ensuring the uniformity of the slurry.

[0045] In a preferred embodiment, the electrically controlled rotating duct consists of an upper rigid duct and an end swing section. The end swing section is hinged to the upper duct via a hydraulic or electric push rod, enabling ±90° pitch or yaw under the drive of a submersible motor. The duct has built-in cables and a communication module. After the ROV is mechanically connected to the duct via a detachable gripper, a wired communication link is established between the two. Directional commands issued by the ROV operator or the automatic decision-making algorithm module are relayed through the ROV and sent directly to the drive controller at the end of the duct via this link, thereby enabling the ROV to remotely control the orientation of the duct. When a release command is executed, the gripper is released, the ROV and the duct are physically separated, and the duct can be retrieved separately.

[0046] The grouting control unit includes an industrial computer, a communication module, a memory storing a three-dimensional scour pit pre-filling model, a real-time data fusion and deviation analysis module, and an automatic decision-making algorithm module. The real-time data fusion and deviation analysis module is used to compare the real-time mud surface elevation data acquired by the underwater sensing unit with the three-dimensional scour pit pre-filling model in the memory point by point to generate a three-dimensional deviation distribution map. The automatic decision-making algorithm module is based on PID control or model predictive control (MPC) algorithm. According to the generated deviation distribution map and the data from the flow velocity and direction sensors, it dynamically generates and outputs control commands for the pumping and execution units, such as adjusting the pumping flow rate of the high-pressure concrete pump, controlling the residence time, swing path, and lifting speed of the electrically controlled rotating duct.

[0047] The unmanned remotely operated vehicle (UAV) is connected to the electrically controlled rotating duct via a detachable clamp, and the UAV is positioned in front of the duct for guidance, enabling precise positioning and orientation of the duct tip. This structure ensures the spatial accuracy of the duct tip in complex underwater environments and allows for rapid release in emergency situations.

[0048] In this embodiment, the digital twin platform is a software platform running on the industrial computer of the injection control unit. It integrates a three-dimensional scour pit pre-filling model, a real-time data fusion and deviation analysis module, and an automatic decision-making algorithm module. The platform receives data from the underwater sensing unit through the communication module and sends the generated control commands to the pumping and execution unit. In some distributed architectures, the digital twin platform can also be deployed on a dedicated server of the surface support platform and interact with the underwater equipment through fiber optic or acoustic modems.

[0049] The method for constructing the three-dimensional scour pit prefill model is as follows:

[0050] Step S1-1, Target Terrain Meshing: Import the high-precision digital terrain model (1cm resolution) obtained from the unmanned surface vessel survey in Step S1 into the digital twin platform. Establish a Cartesian global coordinate system (X,Y,Z) with the pile foundation center as the origin. Perform voxelization on the scour pit area, setting the voxel grid size to 10cm×10cm×1cm. Set the overfill coefficient to 1.05, and linearly interpolate the initial seabed elevation Z_initial and the design repair elevation Z_design of each grid point (X_i,Y_j) to obtain the target elevation Z_target(X_i,Y_j) of that point. The set of all target elevations constitutes the target terrain layer of the prefilled model.

[0051] Step S1-2, Layered Slicing: Based on the underwater stacking stability of the solidified soil material, the total filling thickness is divided into 3-5 grouting layers, with each layer set to a thickness of 0.5m-1.0m. The target terrain layer is horizontally sliced ​​according to its layer height to generate the intermediate target elevation surface after each layer is grouted.

[0052] Steps S1-3, Path Point Generation and Attribute Assignment: On each slice, starting from the deepest point of that layer, a series of infusion path points are generated using a spiral algorithm. Each path point P_k contains the following attributes: three-dimensional coordinates (X_k, Y_k, Z_k), the expected elevation value H_exp, the expected residence time T_exp (initially set to 2 seconds), and the initial value of the duct swing angle (set to 0°). All path points and their attribute sequences together constitute the infusion strategy layer of the three-dimensional scour pit prefilling model.

[0053] Steps S1-4, Model Storage: The target terrain layer and the irrigation strategy layer are stored in the memory in the form of a structured data table, and are called by the real-time data fusion and deviation analysis module through the API interface to realize point-by-point comparison by coordinate index.

[0054] In this embodiment, the electrically controlled rotating duct consists of an upper rigid duct and an end swing section. The end swing section is hinged to the upper duct via a hydraulic or electric push rod, and can achieve pitch and yaw under the drive of a submersible motor. The duct integrates a communication module and a drive controller.

[0055] The unmanned remotely operated vehicle (UAV) is mechanically connected to the conduit via a detachable clamp, and a wired communication link is established simultaneously. When remote control of the conduit's movement is required, directional commands issued by the industrial control computer on the UAV or the surface operator are transmitted directly to the drive controller at the end of the conduit via this wired link. In the event of an emergency (such as conduit jamming or UAV displacement), the infusion control unit issues a release command, triggering the solenoid valve within the clamp to physically separate the UAV from the conduit. The conduit can then be retrieved independently, and the UAV can autonomously avoid obstacles or surface.

[0056] As a specific example of control logic, when the deviation distribution map shows that the elevation of a specific injection point A is lower than the expected value Δh, the Model Predictive Control (MPC) module will estimate the elevation increment that the conduit can bring by increasing the residence time Δt at that point based on parameters such as the location of the point, the current flow velocity v and direction θ, and the slurry consistency, through its internal predictive model. It will then output a command to extend the residence time of the conduit at point A by Δt. At the same time, if the flow velocity and direction sensor detects a transverse flow, the MPC will calculate online the conduit swing angle α required to shift the slurry landing point in the opposite direction, and command the conduit to swing α angle in the opposite direction of the ocean current to achieve dynamic compensation.

[0057] The Model Predictive Control (MPC) in the automatic decision-making algorithm module is designed as follows:

[0058] 1. Predictive Model: A discrete-time state-space model is established based on the principle of mass conservation and semi-empirical formulas. The state variable is defined as x(k) = [e(k), θ(k)]^T, where e(k) is the elevation deviation of the current injection point (actual elevation - target elevation, unit: m), and θ(k) is the slurry accumulation slope (height-to-width ratio, dimensionless). The control variable is defined as u(k) = [v(k), α(k)]^T, where v(k) is the duct lifting speed (unit: m / min), and α(k) is the horizontal oscillation angle of the duct (unit: °, a positive value indicates oscillation in the opposite direction to the ocean current). The predictive model is expressed as:

[0059]

[0060]

[0061] Where Ts is the control cycle (taken as 1 second), q_in(k) is the pump flow rate (m³ / min), A is the diffusion area at the end of the conduit (m²), and k1 and k2 are empirical coefficients calibrated through water tank tests (k1=0.05, k2=0.02).

[0062] 2. Optimize the objective function: Within the prediction time domain Np (10 seconds) and the control time domain Nc (3 seconds), minimize the following cost function:

[0063]

[0064] Where Q, R, and S are weighting matrices (Q=10, R=[1,0;0,1], S=[0.1,0;0,0.5]), ρ is the weighting coefficient (ρ=0.01), and Δu is the rate of change of the control increment, used to suppress violent jitter of the control action.

[0065] 3. The constraints are as follows;

[0066] Catheter lifting speed: 0.1 m / min ≤ v(k) ≤ 0.8 m / min

[0067] Swing angle: -90°≤α(k)≤90°

[0068] Accumulation slope: 1:15≤θ(k)≤1:10

[0069] Elevation deviation: -0.05m≤e(k)≤0.10m (underfill tolerance is greater than overfill tolerance)

[0070] 4. Solution method: In each control cycle, the above constrained quadratic programming problem is transformed into the standard QP form and solved online using the effective set method to obtain the optimal control sequence u*(k). Only the first control command (v(k), α(k)) is output to the pumping and execution unit.

[0071] Furthermore, in one embodiment, the control instructions generated by the automatic decision-making algorithm module for the pumping and execution unit also include instructions to the high-pressure concrete pump to perform reverse pumping operation to handle pipe blockage events. This design achieves rapid and autonomous handling of pipe blockage events through an automatic reverse unblocking mechanism, effectively reducing manual intervention and construction interruption time, and lowering the risk of pipe bursts.

[0072] Furthermore, in one embodiment, the injection control unit is also configured to automatically switch to using data from a mud level gauge to control the lifting height of the electrically controlled rotating duct when the multibeam imaging sonar or profiling sonar fails due to water turbidity. This feature utilizes sensor redundancy design to solve the problem of sonar sensing failure in turbid water, ensuring that the injection process can continue reliably under harsh water quality conditions, and significantly enhancing the system's environmental adaptability.

[0073] Method implementation, for example Figure 1 As shown, the present invention also provides a method for intelligent grouting of solidified soil for offshore wind turbine pile foundations using the above system. The method will be described in detail below with reference to specific embodiments.

[0074] Example 1: Basic implementation process. In a certain offshore wind farm, a scour pit with a maximum depth of 5.2 meters has been formed around the single pile foundation.

[0075] Before construction, the planning work in step S1 is carried out first. The operators use an unmanned surface vessel equipped with a multibeam echo sounder to conduct a full-coverage scan of the scour pit around the pile foundation. The resolution of the high-precision digital terrain model is set to 1 cm, which accurately obtains the three-dimensional shape of the scour pit. Then, the data is imported into the digital twin platform on the shore or ship. The platform divides the scour pit into 4 horizontal grouting blocks from the deepest point upwards according to the design filling volume and sets an overfilling coefficient of 1.05 to compensate for material loss. Then, it automatically plans a grouting path that starts from the deepest point and extends outwards in a spiral, ensuring that the solidified soil can densely fill the entire pit from bottom to top and from the inside to the outside.

[0076] Subsequently, the system proceeds to step S2, which involves lowering and initial positioning. The ROV and the electrically controlled rotating guide tube are connected together via a detachable clamp and then lowered into the water. During the lowering process, the ROV transmits topographic maps and forward video from multi-beam imaging sonar scans in real time. The operator or the automatic decision-making module guides the end of the guide tube to safely approach the bottom of the pit. When it reaches the predetermined height above the bottom of the pit, the mud level gauge accurately determines the distance and automatically adjusts the end of the guide tube to the initial injection height of 0.5 meters from the bottom of the pit.

[0077] Next, the system proceeds to step S3, which involves layered adaptive grouting. The initial pumping flow rate for the first layer is calculated based on the ratio of the designed volume to the expected grouting time to prevent the initial flow rate from being too high or too low. During grouting, multibeam imaging sonar continuously scans at a high frequency of one frame every 3 seconds to acquire the actual diffusion pattern and accumulation height of the slurry underwater in real time. The digital twin platform compares the actual filling elevation with the expected design elevation point by point. When the platform detects that the filling height of a certain area is 30mm lower than the design value, the system automatically decides to increase the residence time of the conduit above that area. When the system detects that the slurry is shifted to the northwest due to the influence of the ocean current (flow velocity 0.3m / s, direction southeast), the automatic decision-making algorithm module instructs the end of the electrically controlled rotating conduit to swing in the southeast direction (i.e., the opposite direction of the ocean current) to compensate for the slurry shift. When the accumulation height reaches the preset elevation of the layer, the system prompts and automatically controls the conduit to move to the next grouting point. After each layer of grouting is completed, the conduit is raised to a certain height, and the above process is repeated.

[0078] During the grouting process, the abnormal handling step S4 is executed simultaneously. The pressure sensor monitors the pump pressure in real time. When a sudden increase in pump pressure exceeding 20% ​​and a simultaneous decrease in flow rate exceeding 15% are detected, the system determines it to be a pipe blockage event. The automatic decision-making algorithm module immediately instructs the high-pressure concrete pump to perform a reverse pumping operation for 4 seconds. If the pressure returns to normal, grouting continues; if the fault is not resolved, the system automatically alarms and stops grouting, awaiting manual intervention. At the same time, multi-beam imaging sonar detects the grout diffusion radius in real time. If the diffusion radius is found to be greater than the design value (e.g., the design value is 2 meters), the system will detect the grout diffusion radius. (The measured value was 4.2 meters, more than twice the normal value). The system determined this to be an event of excessive slurry dispersion. At this time, the system automatically reduced the pumping pressure and instructed the ship's mixing system to increase the dosage of thickening admixtures to increase the slurry viscosity from the initial 500 mPa·s to 800 mPa·s to resist ocean current dispersion. In addition, the system continuously monitored the connection status between the ROV and the conduit. Once it detected that the detachable clamp had accidentally detached or the positioning signal had been lost, the system determined this to be an ROV displacement event, immediately and automatically stopped pumping, and activated the acoustic beacon to guide the recovery of the ROV.

[0079] After the entire pile foundation is poured, the process proceeds to step S5, which involves post-inspection and supplementary grouting. Without the need for equipment retrieval, an ROV equipped with multi-beam imaging sonar performs a 360° scan around the pile foundation. The generated post-fill terrain model is compared with the preset final design terrain model. The system automatically calculates the total volume of the unfilled area to be 1.2 cubic meters, accounting for 4% (less than 5%) of the total filling volume. Since the error exceeds the preset threshold of 3% (the actual error is 4.2%), the system automatically decides to perform supplementary grouting. Given the small amount of supplementary grouting, the system directly controls the duct to quickly return to the underfilled target area for secondary grouting without needing to replan the global path, significantly improving operational efficiency.

[0080] Example 2: Catheter lifting speed control based on neural network. This example optimizes the catheter lifting speed control in step S3 based on Example 1. In order to solve the problem of unstable accumulation slope caused by manual experience in controlling the lifting speed, this invention adopts a regression prediction model based on BP (backpropagation) neural network to determine the catheter lifting speed in real time.

[0081] The input layer nodes of this BP neural network model correspond to three key parameters: the real-time pumping flow rate of the high-pressure concrete pump (unit: cubic meters / hour), the slurry consistency measured by the mud densitometer (unit: seconds, representing fluidity), and the current accumulation slope (aspect ratio) measured by the profile sonar. The output layer node is a single duct lifting speed (unit: centimeters / minute). The model is trained offline using historical grouting data. The core constraint during training is to strictly control the accumulation slope of the solidified soil underwater between 1:10 and 1:15. This is the optimal slope range that has been experimentally verified to ensure slope stability and achieve efficient filling.

[0082] In actual grouting, when the pumping flow rate is 45 cubic meters per hour, the slurry consistency is 12 seconds, and the current accumulation slope is 1:12, the industrial control computer inputs these data into the trained BP neural network model. After the model performs rapid calculations, it outputs an optimal lifting speed, such as 25 centimeters per minute. Compared with fixed speed or control based on simple rules, this control method based on artificial intelligence models can more accurately maintain the ideal accumulation shape, effectively preventing slope collapse caused by lifting too fast or excessive local accumulation of slurry caused by lifting too slow, and significantly improving the overall quality of the filling material.

[0083] The specific structure and training method of the BP neural network are as follows:

[0084] Acquisition of the S1 training dataset: In a 1:10 scale physical model test of a water tank, a pile foundation scour pit model similar to the actual site was constructed. The pumping flow rate Q was set to vary within the range of 30–80 m³ / h (step size 5), the slurry consistency C (expressed as slump spread) within the range of 400–800 mm (step size 50), and the initial accumulation slope S (aspect ratio) within the range of 1:20–1:5. For each set of (Q, C, S) parameters, different guide pipe lifting speeds V (0.1–0.9 m / min, step size 0.05) were tested using a grid search method. The optimal V value was recorded when the final accumulation slope stabilized between 1:10 and 1:15 with no significant slope collapse. A total of 2500 valid samples were collected and randomly divided into training, validation, and test sets (80%, 10%, and 10% respectively).

[0085] S2 data preprocessing: Min-Max normalization is performed on all input variables, mapping them to the [0,1] interval. The transformation formula is: x_norm=(x-x_min) / (x_max-x_min).

[0086] The S3 network structure employs a 3-layer backpropagation (BP) neural network. The input layer has 3 nodes, corresponding to (Q_norm, C_norm, S_norm). The number of hidden layer nodes was determined to be 12 through 5-fold cross-validation (testing with 5-20 nodes, 12 nodes resulted in the smallest mean squared error in the validation set). The hidden layer activation function is the Sigmoid function: f(x) = 1 / (1 + e^{-x}). The output layer has 1 node, corresponding to V_norm, and the activation function is the Purelin linear function.

[0087] S4 training parameters: learning rate set to 0.01, momentum factor set to 0.9, target mean squared error (MSE) set to 1e-4, maximum number of training epochs set to 1000. The Levenberg-Marquardt algorithm is used for training (due to its fast convergence speed), and the loss function is the mean squared error: MSE=(1 / n)Σ(V_pred-V_true)².

[0088] S5 Training and Validation: Iterative training was performed using the training set, with evaluation on the validation set every 100 epochs. Training was terminated early to prevent overfitting when the MSE on the validation set stopped decreasing for 10 consecutive epochs. The final MSE obtained on the test set was 8.7e-5, and the coefficient of determination R² was 0.96, meeting the requirements for engineering applications. The network weights and biases were then stored in the memory of the infusion control unit after training.

[0089] S6 Online Usage: During actual grouting, the system collects real-time pump flow rate, slurry consistency, and current accumulation slope measured by profile sonar every 0.5 seconds. After normalization, the data is input into the solidification network, forward propagation is used to calculate the output value, and then inverse normalization is performed to obtain the real-time optimal guide pipe lifting speed command.

[0090] Example 3: Sensor redundancy and recharge tiered decision-making. This example further illustrates the sensor failure and recharge strategy in complex underwater environments.

[0091] During the injection of a certain layer in step S3, due to the disturbance of the previously injected slurry, the local area of ​​the water becomes extremely turbid. At this time, the echo signal of the multibeam imaging sonar is severely attenuated, causing data failure. The system monitors the data quality in real time. Once the sonar data is detected as unreliable, the automatic decision-making algorithm module immediately executes the sensor redundancy switching strategy: stop using the data from the multibeam imaging sonar, and instead rely entirely on the data from the mud level gauge to control the lifting height of the guide pipe. At the same time, the system continues to use the profile sonar, which is not sensitive to turbidity, to monitor the accumulation slope. Through this redundancy design, the injection process can still be carried out continuously and reliably under extremely harsh visual conditions.

[0092] In the post-detection and refilling stage of step S5, if the system calculates that the total volume of the unfilled area is large, for example, 15 cubic meters, accounting for 15% of the total filling volume, which is much greater than the 3% threshold, the system will not perform simple rapid fixed-point refilling. Instead, based on the hierarchical decision logic, it will determine that the refilling amount (15%) is greater than 5% of the total filling volume, which is a large-scale underfilling. The system will judge this situation as a major defect and automatically decide to execute a complete secondary refilling process: re-call the original scour pit model in the memory, combine it with the currently filled terrain, generate a new supplementary filling layer model and irrigation path, and start the layered adaptive irrigation again from step S3. This hierarchical refilling strategy ensures the repair quality while avoiding the problem of weak filling interface that may be caused by local fixed-point irrigation when large-area filling is required, thus achieving a balance between efficiency and quality.

[0093] In summary, the offshore wind turbine pile foundation solidification soil grouting system and method provided by this invention, through the deep integration of underwater precise sensing, intelligent decision-making algorithms, adaptive execution mechanisms and comprehensive anomaly handling mechanisms, realizes closed-loop intelligent control of the entire process of scour pit repair operations, significantly improving construction quality, efficiency and safety, and has extremely high engineering application value.

[0094] It should be noted that, in this document, relational terms such as first and second are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. The contents not described in detail in this specification are prior art known to those skilled in the art.

[0095] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Those skilled in the art can readily implement the present invention based on the accompanying drawings and the above description. However, any modifications, alterations, or variations made by those skilled in the art without departing from the scope of the present invention, utilizing the disclosed technical content, are equivalent embodiments of the present invention. Furthermore, any modifications, alterations, or variations made to the above embodiments based on the essential technology of the present invention are still within the protection scope of the present invention.

Claims

1. A soil grouting system for offshore wind turbine pile foundations, comprising an underwater sensing unit, a pumping and execution unit, and a grouting control unit, characterized in that, The underwater sensing unit includes an unmanned remotely operated vehicle (UAV) and a multibeam imaging sonar, a profile sonar, and a mud level height gauge mounted on the UAV, as well as a flow velocity and direction sensor installed near the duct opening. The pumping and execution unit includes a high-pressure concrete pump, an electrically controlled rotary duct, and a pressure sensor and a mud density meter installed inside the duct. The end of the electrically controlled rotary duct can swing ±90°, and its orientation can be remotely controlled by an unmanned remotely operated vehicle. The grouting control unit includes an industrial computer, a communication module, a memory storing a three-dimensional scour pit pre-filling model, a real-time data fusion and deviation analysis module, and an automatic decision-making algorithm module; The three-dimensional scour pit prefilling model includes target terrain grid data and injection path point attribute data established in a global coordinate system; The real-time data fusion and deviation analysis module is used to uniformly convert the real-time mud surface elevation data obtained by the underwater sensing unit to the global coordinate system, and then compare it with the target terrain grid data in the three-dimensional scour pit prefill model point by point according to the same horizontal coordinate to generate a three-dimensional deviation distribution map. The automatic decision-making algorithm module, based on the three-dimensional deviation distribution map and the flow velocity and direction data collected by the flow velocity and direction sensor, uses PID control or model predictive control to generate and output control commands for the pumping and execution units, so as to adjust the residence time, swing direction and lifting speed of the electrically controlled rotating duct. The unmanned remotely operated vehicle (UAV) is connected to the electrically controlled rotating conduit via a detachable clamp, and the UAV is positioned on the horizontal projection plane in front of the conduit tip along the infusion direction to achieve precise positioning and orientation of the conduit tip.

2. The offshore wind turbine pile foundation solidification soil grouting system according to claim 1, characterized in that, The automatic decision-making algorithm module generates control commands for the pumping and execution units, including commands to the high-pressure concrete pump to perform reverse pumping operations to handle pipe blockage events.

3. The offshore wind turbine pile foundation solidification soil grouting system according to claim 1, characterized in that, The injection control unit is also configured to automatically switch to using data from a mud level gauge to control the lifting height of the electrically controlled rotating guide tube when the multibeam imaging sonar or the profiling sonar fails due to water turbidity.

4. A method for intelligent grouting of solidified soil for offshore wind turbine pile foundations using the system described in claim 1, characterized in that, Includes the following steps: Step S1: Before construction, use an unmanned boat equipped with a multibeam echo sounder to conduct a full-coverage survey of the scour pits around the pile foundation, generate a high-precision digital terrain model, and divide the grouting blocks in the digital twin platform according to the design filling volume, and plan the grouting path starting from the deepest point and extending outward in a spiral shape. Step S2: Connect the unmanned remotely operated vehicle (UAV) and the electrically controlled rotating guide tube through a detachable clamp and submerge them in the water. The UAV acquires underwater terrain and video in real time, and the end of the guide tube automatically adjusts to the initial height from the bottom of the pit. Step S3 involves layered grouting. During each layer of grouting, the multibeam imaging sonar continuously scans to acquire the actual diffusion pattern of the grout. The digital twin platform compares the actual filling elevation with the expected design elevation and automatically executes one or more of the following adaptive controls based on the comparison deviation: Adjust the dwell time or oscillation speed of the catheter in a specific area; The instruction states that the duct should swing in the opposite direction of the ocean current to compensate for the slurry deviation; When the accumulation height reaches the set value, a prompt will appear indicating to move to the next injection point; Step S4: During the injection process, abnormal situations are monitored in real time and handled automatically. The abnormal situations include pipe blockage events caused by a sudden increase in pump pressure and a decrease in flow rate, excessive dispersion of slurry, and loss of position of unmanned remotely operated vehicles. Step S5: After the filling is completed, the unmanned remotely operated vehicle and multibeam imaging sonar are used to perform a 360° scan to generate a post-fill terrain model. The volume of the unfilled area is automatically calculated. If the error exceeds the threshold, a decision is made and supplementary filling is performed.

5. The method according to claim 4, characterized in that, In step S3, the lifting speed of the electrically controlled rotating duct is determined in real time by a regression prediction model based on a BP neural network. Its input layer nodes correspond to the pump flow rate, slurry consistency, and current accumulation slope, while the output layer nodes correspond to the duct lifting speed. The constraint target during model training is to control the accumulation slope between 1:10 and 1:

15.

6. The method according to claim 4, characterized in that, Step S3 further includes: when the multibeam imaging sonar data fails due to slurry turbidity, automatically switching to the data from the mud level gauge to control the lifting height of the guide tube.

7. The method according to claim 4, characterized in that, In step S4, the automatic handling of the pipe blockage event is as follows: when a sudden increase in pump pressure exceeding 20% ​​and a simultaneous decrease in flow rate exceeding 15% are detected, the high-pressure concrete pump is instructed to perform a reverse pumping operation for 3-5 seconds; if the fault is not resolved after reversal, an alarm is automatically triggered and grouting is stopped. The automatic handling of the slurry over-dispersion event is as follows: when the sonar detects that the diffusion radius of the slurry underwater is greater than twice the design value, the pumping pressure is automatically reduced and the shipboard mixing system is instructed to increase the amount of admixture to increase the viscosity of the slurry. The automatic handling of the unmanned remotely operated vehicle (UAV) dislocation event is as follows: when it is detected that the UAV has detached from the duct or the positioning signal has been lost, the pumping is automatically suspended and the acoustic beacon is activated to guide the recovery.

8. The method according to claim 4, characterized in that, In step S5, if the calculated volume error of the unfilled area is greater than 3%, the system will automatically decide to refill; if the refill amount is less than 5% of the total filling volume, the system will directly control the conduit to return to the target area for secondary refilling without having to replan the global path.

9. The method according to claim 4, characterized in that, In step S1, the high-precision digital terrain model has a resolution of centimeters, the design filling amount includes an overfill coefficient, and the grouting block is divided into multiple layers according to the depth of the scour pit.

10. The method according to claim 4, characterized in that, In step S3, the initial flow rate of each layer of perfusion is calculated as the ratio of the designed volume to the expected perfusion time, and the scanning frequency of the multibeam imaging sonar is a high-frequency periodic scan.

Images