A construction method for an ultra-large land caisson passing through an ultra-thick clay layer

Abstract

The present application relates to the field of bridge caisson construction and discloses a construction method for an ultra-large land caisson penetrating an ultra-thick clay layer, the method comprising: constructing and dynamically updating the digital twin model using real-time perception data streams; running a posture evolution prediction model and a ternary equilibrium decision model based on the model to generate optimal collaborative operation instructions including active preventive corrections; driving a collaborative operation execution system to complete the instructions, forming a closed-loop intelligent control; the system comprises: a collaborative operation execution system that integrates mechanical soil extraction and hydraulic auxiliary equipment; a multi-dimensional real-time perception system that uses sensors such as underwater drones to obtain real-time data on underground and structural status; and an intelligent decision-making system that constructs and runs a digital twin model that interacts with the physical site in real time to make predictive decisions. The present invention significantly improves the efficiency, accuracy, and safety of ultra-large caisson sinking construction through predictive control and intelligent collaborative operations.

Description

Technical Field

[0001] The present invention relates to the field of bridge caisson construction, and in particular to a construction method for an ultra-large land caisson penetrating an ultra-thick clay layer. Background Art

[0002] Caisson construction, a key construction technique for large underground structures, is widely used in projects such as bridge foundations and pump stations. However, when ultra-large land-based caissons need to penetrate extremely thick, highly viscous clay layers, traditional construction methods expose their inherent limitations.

[0003] First, the immense sidewall friction and excavation resistance of the ultra-thick clay layer significantly restricts the efficiency of the caisson's lowering. Conventional mechanical grab buckets, operating blindly underwater, struggle to cope with the clay's strong adhesion, often leading to low soil extraction efficiency and even stagnation of the caisson's sinking.

[0004] More critically, traditional methods suffer from serious flaws in attitude control. Due to the inability to accurately and real-timely monitor the excavation status of the wellbore bottom, particularly beneath the partition walls and blade feet, construction decisions rely heavily on the operator's indirect judgment and subjective experience. This open-loop, lagging control model is highly susceptible to tilting and twisting of the caisson due to localized over- or under-excavation, and can even induce sudden subsidence, posing significant risks to structural safety.

[0005] Therefore, the present invention proposes a construction method for an ultra-large land caisson penetrating an ultra-thick clay layer to address the deficiencies of the prior art. Summary of the Invention

[0006] In response to the shortcomings of the existing technology, the present invention provides a construction method for an ultra-large land caisson to pass through an ultra-thick clay layer, which solves the problems faced by ultra-large land caissons when passing through an ultra-thick clay layer, such as low sinking efficiency and difficulty in posture control due to the high viscosity of the soil, as well as high structural and construction risks caused by many invisible and uncontrollable factors in the construction process.

[0007] To address the above technical issues, the first aspect of the present invention provides a method for constructing an ultra-large land caisson through an ultra-thick clay layer. This method constructs a digital twin that interacts with the construction site in real time to dynamically perceive, accurately predict, intelligently decide, and collaboratively control the entire caisson sinking process. The method includes the following steps:

[0008] First, an initial digital twin model is constructed. This model is a comprehensive digital body that integrates multi-source static information. Among them, geological information includes soil layer distribution data obtained through surveys, soil physical and mechanical parameters, and the maximum allowable drainage excavation depth of the clay layer calculated by establishing a mathematical model based on historical hydrological data. The caisson structure information is a finite element model of the caisson geometry and structure established based on the design drawings. The caisson structure consists of a steel shell concrete caisson section and a concrete caisson section with wedge-shaped blades from bottom to top. The interior is equipped with a well wall, partition walls and partition walls to form a stable force system. The sensor layout information defines the spatial position of various sensors, specifically:

[0009] The placement of mechanical sensors on the grab bucket for acquiring physical and mechanical data;

[0010] The placement of the online slurry property analyzer on the slurry pipeline for obtaining material removal data;

[0011] As well as the key node positions of structural stress sensors and attitude sensors inside the caisson structure.

[0012] Then, during the non-drainage sinking phase, the multi-dimensional real-time sensing system deployed at the construction site is activated to continuously collect data on the construction process, forming a real-time data stream. The specific process is as follows:

[0013] An underwater drone equipped with a multi-beam sonar is used to scan the soil sampling area within the wellbore to generate geometric data representing the three-dimensional shape of the well bottom.

[0014] The mechanical sensors installed on the grab bucket monitor the operating force during the grabbing and lifting process in real time to generate physical and mechanical data.

[0015] An online slurry property analyzer is used to monitor slurry parameters such as density and flow rate to generate material removal data.

[0016] The initial digital twin model is dynamically updated using the above real-time data stream to obtain a real-time digital twin model that can reflect the status of the construction site in real time.

[0017] During the dynamic update process, a key step is that the intelligent decision-making system extracts physical and mechanical data (such as the grab bucket operating force value) from the real-time data stream as input, and uses an inversion analysis algorithm to generate a soil mechanical property distribution map that can characterize the softness and hardness of the underwater soil. This distribution map is then incorporated into the digital twin model in real time to provide a basis for subsequent soil excavation decisions.

[0018] Next, the intelligent decision-making system performs calculations and makes decisions based on the real-time digital twin model. This step includes two core models:

[0019] One of them is the posture evolution prediction model. The function of this model can be represented by the following formula:

[0020] ;

[0021] in, is the predicted caisson posture vector at the next moment; Represents a complex prediction function based on the principles of soil mechanics and rigid body dynamics; is the caisson posture vector at the current moment; A coordinated sequence of actions to be executed for a plan; is the current real-time digital twin model. This model takes the current posture, planned action, and real-time model as input, and calculates and outputs the predicted caisson posture at the next moment.

[0022] The second is a ternary balance decision model. This model uses the predicted caisson posture output by the posture evolution prediction model as an important input and obtains the optimal collaborative operation instructions by solving a comprehensive objective function. This objective function aims to balance the three goals of sinking efficiency, construction risk, and energy consumption. Its form can be expressed as:

[0023] ;

[0024] Among them, the comprehensive objective function Defined as:

[0025] ;

[0026] in, For the optimal collaborative operation instructions; Represents the set of all possible collaborative job instructions; Represents from the set A specific collaborative operation instruction is taken out as the optimization variable; is the comprehensive objective function. is the expected sinking efficiency term; is the comprehensive energy consumption item; It is a comprehensive construction risk item; , , is the weight coefficient corresponding to each item. In particular, the comprehensive construction risk item The calculation couples the deviation degree calculated by the predicted caisson posture with the predicted structural stress level to ensure the safety of decision-making.

[0027] Subsequently, the collaborative operation execution system synchronously controls the mechanical earth-excavation equipment, hydraulic auxiliary equipment, and deviation correction equipment to perform collaborative operations based on the received optimal collaborative operation instructions. Specifically, the intelligent decision-making system first determines the soil properties of the target earth-excavation area based on the soil mechanical property distribution map in the real-time digital twin model. If the area is determined to be hard soil, an instruction is generated to control the grab bucket to perform earth-excavation operations while prioritizing the stirring intensity of the high-pressure water gun corresponding to this area to assist the grab bucket in breaking up the soil. The suction speed of the mud pump is also adjusted in conjunction to match the mud generation rate generated by the collaborative operation of the grab bucket and high-pressure water gun.

[0028] In addition, the method also includes an active preventive correction mechanism. If the intelligent decision-making system runs the posture evolution prediction model and finds that the predicted caisson posture deviates from the preset target threshold, it will incorporate active preventive correction instructions when generating the optimal collaborative operation instructions. The instruction is generated as follows: first, based on the predicted caisson posture, the predicted posture deviation direction and deviation amount are calculated; then, the opposite side of the well wall of the deviation direction is determined as the correction operation area; finally, an instruction is generated to specify the activation of the water jet pipe and air curtain pre-buried in the well wall within the correction operation area, and the operation intensity of the water jet pipe and air curtain is set according to the calculated deviation amount. In this way, before the deviation actually occurs, the foreseeable posture deviation can be hedged in advance by reducing the friction resistance of the opposite well wall.

[0029] A second aspect of the present invention provides a system for constructing an ultra-large land caisson through an ultra-thick clay layer, the system being configured to execute any of the aforementioned methods. The system comprises:

[0030] Multi-dimensional real-time perception system;

[0031] Collaborative job execution system;

[0032] An intelligent decision-making system is communicatively connected with the above two systems.

[0033] The multi-dimensional real-time perception system is responsible for continuously collecting construction site data and forming a real-time data stream.

[0034] The collaborative operation execution system includes mechanical earth-taking equipment, hydraulic auxiliary equipment and correction equipment. The correction equipment includes a water jet pipe and an air curtain pre-buried in the wall of the caisson structure. In a specific embodiment, the system also includes an intelligent earth-taking system. The physical structure of the intelligent earth-taking system is as follows:

[0035] On the shaft wall and the top of the partition wall of the caisson structure, large steel distribution beams, Bailey beams, small steel distribution beams and P50 rails are laid in sequence from bottom to top, and a gantry crane for lifting mud pumps is installed on the P50 rails.

[0036] The intelligent decision-making system is configured to:

[0037] Receive real-time data streams collected by a multi-dimensional real-time perception system and dynamically update the initial digital twin model accordingly to obtain a real-time digital twin model;

[0038] Based on this real-time digital twin model, the posture evolution prediction model and the ternary balance decision model are run to perform multi-objective balance calculations and autonomously generate optimal collaborative operation instructions;

[0039] When the predicted caisson posture is detected to be deviating, proactive preventive correction instructions are incorporated into the optimal collaborative operation instructions;

[0040] The optimal collaborative operation instructions finally generated are sent to the collaborative operation execution system to achieve synchronous control and collaborative operation of mechanical earth-moving equipment, hydraulic auxiliary equipment and correction equipment.

[0041] The present invention provides a construction method for an ultra-large land caisson penetrating an ultra-thick clay layer. It has the following beneficial effects:

[0042] 1. By constructing a digital twin model that interacts with the construction site in real time, this invention transforms the caisson sinking process from a black-box operation reliant on experience into a transparent process capable of quantifiable analysis and scientific deduction. This model deeply integrates geological, structural, and multi-source real-time sensory data, enabling the intelligent decision-making system to perform global, multi-objective balance calculations. This fundamentally changes the traditional construction model of decentralized decision-making based on local data and personnel experience, significantly improving the scientific nature of construction decision-making and the level of comprehensive management and control.

[0043] 2. This invention addresses the challenges of traditional construction methods, such as the inability to observe the underground working surface and the severe lag in attitude control. By utilizing underwater drones and other equipment to build a multi-dimensional real-time perception system, this system overcomes blind spots in visual monitoring of critical areas, such as below the blade foot. More importantly, this invention uses an attitude evolution prediction model to predict the future attitude of the caisson and initiates proactive, preventive corrections. This fundamental shift from passive lag correction to proactive, advanced control significantly improves the accuracy and timeliness of attitude control, providing a key guarantee for the safe sinking of ultra-large caissons.

[0044] 3. This invention uses real-time inversion to generate a distribution map of soil mechanical properties, accurately identifying the softness and hardness of underwater soil and adaptively controlling the operating parameters of the grab bucket, high-pressure water gun, and mud pump accordingly. This precise, intelligent collaborative operation effectively overcomes the technical bottleneck of low earth extraction efficiency caused by the high viscosity of the soil when ultra-large caissons penetrate extremely thick clay layers, achieving efficient and continuous earth removal.

[0045] 4. This invention integrates real-time perception, predictive decision-making, and coordinated execution into an organic, closed-loop intelligent control system. This system systematically addresses the complex technical challenges of coupling sinking efficiency, attitude stability, and structural risk during the sinking of ultra-large land caissons. This system reduces the construction process's reliance on subjective experience, achieving standardized and refined construction. While ensuring a stable and controllable process, it also provides reliable technical support for managing complex geological conditions, effectively reducing overall project risk. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 It is a schematic diagram of the overall arrangement of the present invention;

[0047] Figure 2 A top view of the overall arrangement of the present invention;

[0048] Figure 3 It is a three-dimensional cross-sectional view of the caisson structure of the present invention;

[0049] Figure 4 A schematic diagram of the mud processing system of the present invention Figure 1 ;

[0050] Figure 5 A schematic diagram of the mud processing system of the present invention Figure 2 ;

[0051] Figure 6 A schematic diagram of the inner wellbore of the caisson structure of the present invention;

[0052] Figure 7 A schematic diagram of the inner ring wellbore partition wall of the caisson structure of the present invention;

[0053] Figure 8 A schematic diagram of the outer ring well hole of the caisson structure of the present invention;

[0054] Figure 9 Schematic diagram of inner and outer ring wellbore partition walls of the caisson structure of the present invention;

[0055] Figure 10 A schematic diagram of the outer ring wellbore partition wall of the caisson structure of the present invention;

[0056] Figure 11 A schematic diagram of the middle wellbore of the caisson structure of the present invention;

[0057] Figure 12 This is a functional module diagram of the intelligent decision-making system of the present invention;

[0058] Figure 13 This is a schematic diagram of the collaborative operation and active preventive correction principle of the present invention;

[0059] Figure 14It is a flow chart of the construction method of the present invention;

[0060] Figure 15 This is a functional block diagram of the construction system of the present invention;

[0061] Figure 16 This is a schematic diagram of the dynamic update principle of the real-time digital twin model of the present invention.

[0062] Among them: 1. Caisson structure; 11. Steel shell concrete caisson; 12. Concrete caisson; 13. Blade foot; 14. Well wall; 15. Partition wall; 16. Partition wall; 17. Bottom seal concrete; 18. Core fill concrete; 19. Top plate; 2. Dewatering well; 3. Crawler crane; 4. Grab bucket; 5. Small excavator; 6. Tower crane; 7. Mud handling system; 71. Mud pump; 72. High-pressure water gun; 73. Water replenishment tank; 74. High-pressure water Pump; 75. Sedimentation tank; 76. Slurry pipe; 77. High-pressure water supply pipe; 8. Underwater drone; 9. Intelligent soil excavation system; 91. Gantry crane; 92. P50 rail; 93. Small distribution beam; 94. Bailey beam; 95. Large steel distribution beam; 100. Collaborative operation execution system; 110. Multi-dimensional real-time perception system; 120. Intelligent decision-making system; 121. Posture evolution prediction model; 122. Ternary balance decision-making model. DETAILED DESCRIPTION

[0063] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the present specification. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0064] Reference Figure 1 、 Figure 12 、 Figure 13 、 Figure 14 、 Figure 15 and Figure 16 The present invention provides an intelligent construction method and system for constructing an ultra-large land caisson through an ultra-thick clay layer. The system physically comprises a caisson structure 1 to be sunk, and integrates a collaborative operation execution system 100, a multi-dimensional real-time perception system 110, and an intelligent decision-making system 120 as the core control system.

[0065] The embodiment of the present invention provides a method for constructing a super-large land caisson through a super-thick clay layer, referring to Figure 14 , which may include the following steps:

[0066] S1, build an initial digital twin model including geological information, caisson structure information and sensor layout information.

[0067] S2, during the undrained sinking stage, the multi-dimensional real-time perception system 110 deployed at the construction site continuously collects data to form a real-time data stream, and uses the real-time data stream to dynamically update the initial digital twin model to obtain a real-time digital twin model.

[0068] S3, the intelligent decision-making system 120 runs the posture evolution prediction model based on the real-time digital twin model to generate a predicted caisson posture, and runs the ternary balance decision model to perform multi-objective balance calculations based on the predicted caisson posture, and autonomously generates optimal collaborative operation instructions.

[0069] S4, the collaborative operation execution system 100 synchronously controls the mechanical earth-taking equipment, hydraulic auxiliary equipment and correction equipment to perform collaborative operation according to the optimal collaborative operation instruction.

[0070] S5, if the predicted caisson posture deviates from the preset target, an active preventive correction instruction is incorporated into the generation of the optimal collaborative operation instruction, so as to offset the posture deviation indicated by the predicted caisson posture in advance through collaborative operation.

[0071] In step S1, the initial digital twin model is a comprehensive digital entity that integrates multi-source static information. Figure 2 and Figure 3 The caisson structure information is the geometry and structural model of the caisson structure 1 established based on the design drawings. The geological information is the soil layer distribution data and soil physical and mechanical parameters obtained through geological surveys. The sensor layout information defines the three-dimensional spatial positions of various sensors.

[0072] In step S2, refer to Figure 1 and Figure 16 After the multi-dimensional real-time perception system 110 is activated, its multiple internal sensors begin to operate, forming a multi-source heterogeneous real-time data stream. For example, the underwater drone 8 equipped with a multi-beam sonar scans the bottom of the well, generating geometric data; the mechanical sensors deployed on the grab 4 monitor the operating force value and generate physical and mechanical data; the online slurry property analyzer monitors mud parameters and generates material removal data. The intelligent decision-making system 120 receives this data stream and updates the initial digital twin model. This update process includes an inversion analysis step, which uses physical and mechanical data as input to inversely generate a soil mechanical property distribution map that can characterize the softness and hardness of the underwater soil, and then incorporates this distribution map into the real-time digital twin model.

[0073] In step S3, refer to Figure 12 The intelligent decision-making system 120 performs calculations based on the updated real-time digital twin model to generate work instructions. This process is completed by two core models: the posture evolution prediction model 121 and the ternary balance decision model 122.

[0074] In step S4, refer to Figure 4 and Figure 13 After receiving the optimal collaborative operation instruction, the collaborative operation execution system 100 analyzes it and synchronously controls the various devices under its control. For example, if the instruction requires excavation in an area identified as hard soil, the system will synchronously output a control signal to drive the grab bucket 4, increase the stirring intensity of the high-pressure water gun 72 at the corresponding location, and jointly adjust the suction speed of the mud pump 71.

[0075] In step S5, refer to Figure 12 、 Figure 13 and Figure 14 , when the predicted caisson posture vector deviates from the preset target threshold, the intelligent decision-making system 120 starts the active preventive correction process. First, the posture deviation direction and deviation amount indicated by the caisson posture vector are calculated. Then, the opposite side of the well wall of the deviation direction is determined as the correction operation area. Finally, an active preventive correction instruction is incorporated into the generated optimal collaborative operation instruction, which specifies the start-up of the water jet pipe and air curtain pre-buried in the well wall 14 in the correction operation area, and sets its operation intensity and duration according to the deviation amount, so as to offset the posture deviation by reducing the opposite side friction before the deviation actually occurs.

[0076] The specific structures and steps of the construction system and method described in the embodiments of the present invention are described in detail below.

[0077] Reference Figures 1-11 The construction object in this embodiment is a caisson structure 1, which is a large permanent underground structure. The caisson structure 1 is a composite structure in the vertical direction, with a lower portion being a steel shell concrete caisson 11 and an upper portion being one or more sections of concrete caisson 12 connected in height.

[0078] The steel shell concrete caisson 11 is located at the bottom of the structure. Its outer shell is made of steel and concrete is poured inside. At the bottom of the steel shell concrete caisson 11, a wedge-shaped blade foot 13 is provided. The geometric shape of the wedge-shaped blade foot 13 is designed to be inclined outside and straight inside, or inclined inside and outside. Its function is to concentrate stress during the sinking process, cut into the soil, and guide the soil to move into the well or to the inside and outside sides to reduce the sinking resistance.

[0079] In terms of plane layout, refer to Figure 2-Figure 3The caisson structure 1 is composed of an outer well wall 14 and an internal partition wall 15 and partition wall 16, which together form a grid-like force-bearing system. The well wall 14 forms the outer contour of the caisson structure 1. The partition wall 15 and partition wall 16 crisscross the interior of the caisson structure 1, dividing the interior of the caisson into multiple independent wells. This grid-like design greatly enhances the overall rigidity and strength of the caisson structure 1, enabling it to maintain structural stability and integrity under the huge deadweight and uneven soil pressure.

[0080] To control sinking friction and actively correct the caisson's posture during construction, a water jetting system and an air curtain system are pre-buried within the concrete structure of the caisson wall 14 along its outer surface. These pipes are laid out along the caisson's perimeter, with independent control valves installed in sections. When needed, high-pressure water can be injected into specific areas of the water jetting system, or compressed air can be introduced into the air curtain system. This creates a water or air film between the caisson wall 14 and the soil, thereby locally and selectively reducing sidewall friction, assisting sinking or correcting posture deviation.

[0081] After the caisson structure 1 sinks to the designed elevation, subsequent steps will be carried out in sequence, including pouring bottom sealing concrete 17 to seal the bottom of the well hole, then pouring core filling concrete 18 to fill part or all of the well hole according to design requirements, and finally constructing the top plate 19 on the top to form a complete base foundation.

[0082] Reference Figure 1-Figure 4 The collaborative work execution system 100 in this embodiment of the present invention is a physical entity that executes the work instructions generated by the intelligent decision-making system 120. The collaborative work execution system 100 integrates multiple devices to implement functions such as mechanical soil extraction, hydraulic assistance, and posture correction. The specific components of the system are as follows:

[0083] The collaborative work execution system 100 includes a mechanical earth-moving device. In a specific embodiment, referring to Figure 1 The mechanical earth-excavation equipment primarily consists of one or more crawler cranes 3 and a cooperating grab bucket 4. The crawler crane 3 provides the power for lifting, lowering, and rotating, and is used to operate the grab bucket 4. The grab bucket 4 is the primary earth-excavation tool, relying on its own weight to cut into the soil and, by closing the grab bucket, excavates and removes the soil at the bottom of the wellbore.

[0084] The collaborative work execution system 100 also includes hydraulic auxiliary equipment. Figure 4, the equipment is implemented in the form of a modular mud treatment system 7. In a wellbore, there is a mud pump 71 and one or more high-pressure water guns 72. The function of the high-pressure water gun 72 is to spray high-pressure water onto the clay body, and use the impact and scouring effect of the water flow to loosen and mud the dense soil to assist the grab 4 operation or directly form mud. The mud pump 71 is used to suck out the mud formed in the wellbore. The entire mud treatment system 7 constitutes a closed-loop network: a high-pressure water pump 74 draws clean water from the water supply tank 73, and supplies high-pressure water to the high-pressure water guns 72 in each wellbore through a high-pressure water supply pipe 77 arranged along the well wall 14 or the partition wall 15; the mud sucked out by the mud pump 71 is discharged to the sedimentation tank 75 outside the site through the mud pipe 76 for solid-liquid separation treatment.

[0085] The collaborative operation execution system 100 further includes a correction device. The correction device mainly refers to the water jet pipe and air curtain system pre-buried on the outside of the caisson wall 14. Its working principle is that when the intelligent decision-making system 120 issues a correction instruction, the control unit of the collaborative operation execution system 100 will open the valve of the pipeline in a specific area, inject high-pressure water into the water jet pipe in the area or inject compressed air into the air curtain pipeline. This will form a water film or air film with a lubricating effect between the caisson wall 14 and the surrounding soil, thereby significantly reducing the side wall friction resistance in the area. By actively and asymmetrically adjusting the friction resistance in different areas, a correction torque can be generated to resist or correct the tilt of the caisson.

[0086] In a preferred embodiment, the collaborative operation execution system 100 further includes an intelligent soil taking system 9 as its component, which is mainly used to carry and accurately position hydraulic auxiliary equipment. Figure 5 The physical structure of the intelligent soil extraction system 9 is a multi-layer load-bearing and track platform built on the top of the caisson structure 1. The specific construction method is:

[0087] First, a large steel distribution beam 95 is laid on the top of the caisson wall 14 and the partition wall 15. Its function is to evenly transfer the upper load to the concrete structure below.

[0088] Bailey beam 94 is laid on the large steel distribution beam 95 , and small steel distribution beam 93 is laid on the Bailey beam 94 .

[0089] On the uppermost layer, P50 steel rails 92 are laid in a specific direction to form a track.

[0090] One or more gantry cranes 91 are installed on the P50 rails 92 and can move along the rails.

[0091] The main function of the gantry crane 91 is to suspend the mud pump 71. Through the movement of the gantry crane 91, the mud pump 71 can be accurately positioned and hoisted in the wellbore.

[0092] Reference Figure 1 、 Figure 3 and Figure 16 The multi-dimensional real-time perception system 110 is a hardware assembly used in embodiments of the present invention to continuously acquire construction process status information. This multi-dimensional real-time perception system 110 integrates multiple types of sensors for quantitatively monitoring the construction site from different dimensions. The real-time data streams it outputs are the basis for subsequent intelligent decision-making and control. The specific components of this system are as follows:

[0093] The multi-dimensional real-time perception system 110 includes a sensing device for obtaining bottom hole geometric data. In a specific embodiment, the device is one or more underwater drones 8. The underwater drone 8 is equipped with a multi-beam sonar scanning device. During operation, the underwater drone 8 navigates underwater in the wellbore of the caisson structure 1. The multi-beam sonar it carries transmits and receives sound wave signals to the surface of the soil at the bottom of the well, and generates high-density three-dimensional point cloud data by calculating the round-trip time and angle of the sound waves. These data are used to construct a three-dimensional digital elevation model of the bottom of the wellbore in real time, so as to accurately obtain geometric information such as the burial depth below the blade foot 13, the excavation depth of each wellbore, and the flatness of the bottom of the well.

[0094] The multi-dimensional real-time sensing system 110 also includes sensing equipment for acquiring physical and mechanical data. This equipment consists of mechanical sensors deployed on the grab bucket 4 or its lifting cable. These sensors can be tension sensors or strain gauges. Their function is to monitor in real time the changes in force experienced by the grab bucket 4 during its lowering, bottoming-out, closing, and lifting cycles, such as bottoming-out impact force, closing resistance, and bucket lifting load. This physical and mechanical data directly reflects the strength of the interaction between the grab bucket 4 and the soil at the bottom of the well and serves as a key input for subsequent inverse analysis of the soil's mechanical properties.

[0095] Multi-dimensional real-time sensing system 110 further includes a sensor device for acquiring material removal data. This device is an online slurry property analyzer installed on the mud pipe 76 of the mud processing system 7. This analyzer integrates a flow meter and a densitometer, continuously measuring the instantaneous flow rate and density of the slurry flowing through the pipe. By continuously integrating these two parameters, the volume and mass of soil removed from the wellbore per unit time can be accurately calculated, enabling real-time quantitative monitoring of earth removal.

[0096] In addition, the multi-dimensional real-time sensing system 110 also includes structural state sensors arranged in the caisson structure 1. These sensors include:

[0097] Structural stress sensors fixed at key locations of the shaft wall 14 and partition wall 15 are used to monitor the stress and strain of concrete to assess structural safety;

[0098] And the attitude sensor installed on the top of the caisson structure 1, such as a high-precision inclinometer, is used to measure the inclination of the caisson in two horizontal axes and the overall verticality in real time, providing direct measurement data for attitude control.

[0099] Reference Figure 8 and Figure 11 , the intelligent decision system 120 in the embodiment of the present invention is the computing core that performs data processing, model calculation and decision generation.

[0100] The hardware of the intelligent decision-making system 120 is composed of one or more industrial computers or a server cluster consisting of multiple servers. Depending on the specific deployment plan, the hardware can be deployed in a control center at the construction site or in a remote cloud data center.

[0101] The intelligent decision-making system 120 establishes a two-way communication connection with the multi-dimensional real-time perception system 110 and the collaborative task execution system 100 via a data communication network. Specifically, the intelligent decision-making system 120 receives the real-time data stream collected by the multi-dimensional real-time perception system 110 and sends the collaborative task instructions generated through calculation to the collaborative task execution system 100.

[0102] This communication connection can be achieved through one or more communication technologies, such as industrial Ethernet for wired connections with fixed on-site equipment, or wireless communication technologies (such as 5G or Wi-Fi) for data exchange with on-site mobile devices or remote cloud platforms. The purpose of establishing this communication connection is to provide the system with a stable and low-latency data transmission channel to meet real-time control requirements.

[0103] The specific implementation steps of the construction method according to the embodiment of the present invention will be described in detail below.

[0104] Reference Figure 1 、 Figure 4 and Figure 14 In a preferred embodiment, the construction method of the present invention may include a construction preparation and drainage sinking stage before entering the main undrained sinking stage.

[0105] This phase begins with construction preparation work, which includes:

[0106] The construction site was leveled and the foundation reinforced. Historical hydrological data, particularly the highest river water levels during the construction period in recent years, were analyzed to calculate the foundation bearing capacity and base stability, determining the maximum allowable excavation depth H of the clay layer under drainage conditions. Several drainage wells 2 were arranged in a circular pattern around the perimeter of the caisson structure 1. On the side closest to the river, the spacing of the drainage wells 2 was appropriately increased based on the hydraulic gradient calculation results to enhance the drainage effect. Meanwhile, tower cranes 6 and other lifting equipment were installed on the site.

[0107] After the first section of the caisson structure 1 (e.g., the steel-shell concrete caisson 11) has been assembled and concrete poured, and after it has reached its designed strength, the drainage and sinking phase begins. First, all dewatering wells 2 are activated for continuous dewatering, lowering the groundwater level in the construction area and maintaining it at a depth of 0.5 to 1.0 meters below the bottom of the caisson blade 13, creating dry working conditions within the wellbore.

[0108] During this phase, excavation is performed using a combination of a small excavator 5 and a tower crane 6. A small excavator 5 is hoisted into the caisson's borehole to excavate the soil. The excavated clay is loaded into a bucket by the small excavator 5, which is then hoisted out of the well by the tower crane 6 and dumped into a designated sedimentation tank 75 or storage yard.

[0109] Reference Figure 4 To ensure the smooth sinking of the caisson and the controllable posture, the soil excavation operation strictly follows the preset zoning order. A specific six-step soil excavation order is:

[0110] First, excavate the soil of the inner circle well holes, then the soil under the partition wall between the inner circle well holes, then the soil of the outer circle well holes, then the soil under the partition wall between the inner and outer circle well holes, then the soil under the partition wall between the outer circle well holes and their intersection, and finally excavate the soil of the middle well hole and the rounded corner well hole.

[0111] For areas far from the operating radius of the tower crane 6, where lifting is difficult (e.g., intermediate wellbore), hydraulic assistance can be employed. Specifically, after the small excavator 5 loosens the soil, a high-pressure water gun 72 is used to agitate the clay to form a slurry. This slurry is then pumped via a slurry pipe 76 to a sedimentation tank 75 via a slurry pump 71. Once the clay layer has been excavated to the preset maximum depth H, if the caisson structure 1 has not yet reached the design elevation, the soil beneath the partition walls 15 and, if necessary, the soil beneath the partition walls 16 are further removed symmetrically until the caisson is fully sunk or this stage is complete. Thereafter, the small excavator 5 is removed, and drainage is stopped to allow the groundwater level to recover, preparing for the subsequent undrained sinking phase.

[0112] After the optional drainage sinking stage is completed, or directly started, water is injected into the wellbore of the caisson structure 1 to make the water level in the well at least 2 meters higher than the groundwater level outside the site, forming a non-drainage operation condition. Then, the core non-drainage sinking intelligent construction stage of the present invention is entered. Figures 1-16 The specific steps and principles of this stage are explained as follows:

[0113] S1. Construction of the initial digital twin model:

[0114] This step aims to create a comprehensive digital foundation model that describes the initial state of construction.

[0115] First, data such as soil layer distribution, thickness, and physical and mechanical parameters obtained through geological surveys are digitized to create a three-dimensional geological model. Second, the design drawings of the caisson structure 1, including the geometric dimensions, spatial relationships, and material properties of all components, including the steel-shell concrete caisson 11, concrete caisson 12, caisson wall 14, partition wall 15, and partition wall 16, are converted into a three-dimensional structural information model. Finally, the planned placement locations of all sensors in the multi-dimensional real-time perception system 110 (e.g., the initial path points of the underwater drone 8, the three-dimensional coordinates of the structural stress sensors, and the attitude sensors) are registered in a unified coordinate system to form a sensor placement information model. The geological model, structural information model, and sensor placement information model are spatially aligned and data fused, ultimately forming an initial digital twin model that incorporates multi-dimensional attributes such as geometry, physics, and information.

[0116] S2. Real-time perception and dynamic model update:

[0117] After the undrained subsidence begins, the multi-dimensional real-time sensing system 110 starts continuous operation. Figure 1 Various sensors within the grab bucket continuously collect data, generating real-time, multi-source, heterogeneous data streams. For example, underwater drone 8 uses its multi-beam sonar to scan the well bottom, generating a data stream describing the three-dimensional geometry of the well bottom. Mechanical sensors deployed on grab bucket 4 record data such as bottom impact force and closing resistance during the operation cycle, generating a physical and mechanical data stream. An online slurry property analyzer installed on mud pipe 76 measures the flow rate and density of the mud, generating a material removal data stream.

[0118] After receiving these real-time data streams, the intelligent decision-making system 120 dynamically refreshes the initial digital twin model. A special technical process is that the system uses the received physical and mechanical data streams to perform inverse analysis of the mechanical properties of the soil. Specifically, the mechanical data such as the closing resistance of the grab 4 is used as input, and a preset soil mechanics inversion algorithm is used to calculate the mechanical property parameters such as the shear strength and cohesion of the soil at the corresponding position. By spatially interpolating the data obtained in multiple operation cycles, a real-time updated soil mechanical property distribution map that characterizes the softness and hardness of the underwater soil can be generated. This distribution map is then incorporated into the digital twin model, allowing the model to dynamically and accurately reflect the soil state that is not visible on site.

[0119] S3. The internal mechanism of intelligent decision-making:

[0120] Reference Figure 12 , the intelligent decision-making system 120 is based on the real-time digital twin model and uses its two internal core models, namely the posture evolution prediction model 121 and the ternary balance decision model 122, to perform calculations to generate optimal operation instructions.

[0121] The function of the posture evolution prediction model 121 is to deduce the future impact of a planned action sequence on the caisson's posture. This model takes the current caisson's posture, a planned coordinated action sequence, and the current real-time digital twin model, particularly the soil mechanical property distribution map contained therein, as input. Through mechanical simulation calculations, it outputs a predicted caisson posture at the next moment. Its core functionality can be expressed by the following formula:

[0122] ;

[0123] in, is the predicted caisson posture vector at the next moment; Represents a complex prediction function based on the principles of soil mechanics and rigid body dynamics; is the caisson posture vector at the current moment; A coordinated sequence of actions to be executed for a plan; is the current real-time digital twin model. This model takes the current posture, planned action, and real-time model as input, and calculates and outputs the predicted caisson posture at the next moment.

[0124] The ternary equilibrium decision model 122 uses the predicted caisson posture output by the posture evolution prediction model 121 as its key input. It generates optimal collaborative operation instructions by solving a comprehensive objective function encompassing expected sinking efficiency, comprehensive construction risk, and comprehensive energy consumption. This process aims to find an optimal balance between multiple mutually constrained objectives (i.e., speed, stability, and economy). The calculation of the comprehensive construction risk term couples the deviation degree calculated from the predicted caisson posture with the predicted structural stress level, ensuring that the decision-making process anticipates potential risks.

[0125] The decision-making process is a multi-objective optimization problem, whose goal is to find an action that optimizes the comprehensive objective function from all possible action sets. Its mathematical expression is as follows:

[0126] ;

[0127] Among them, the comprehensive objective function Defined as:

[0128] ;

[0129] in, For the optimal collaborative operation instructions; Represents the set of all possible collaborative job instructions; Represents from the set A specific collaborative operation instruction is taken out as the optimization variable; is the comprehensive objective function. is the expected sinking efficiency term; is the comprehensive energy consumption item; It is a comprehensive construction risk item; , , is the weight coefficient corresponding to each item.

[0130] S4. Execution logic of collaborative work:

[0131] Reference Figure 12 After receiving the optimal collaborative operation instruction generated by the intelligent decision-making system 120, the collaborative operation execution system 100 analyzes it and synchronously controls the mechanical earth-taking equipment, hydraulic auxiliary equipment and correction equipment under its jurisdiction to perform operations.

[0132] Take a specific scenario as an example: when the optimal instruction requires excavation of an area that is determined to be hard soil in the real-time digital twin model, the collaborative operation execution system 100 will synchronously perform the following actions:

[0133] Send a control signal to the crawler crane 3 to lower the grab bucket 4 at a specific speed and height, and close it with the calculated optimal force value after touching the bottom;

[0134] Send a control signal to the mud processing system 7 to start and increase the stirring intensity of the high-pressure water gun 72 near the corresponding operation point, so as to assist in crushing the soil with hydraulic action;

[0135] The suction power of the mud pump 71 in the wellbore is increased in conjunction with the pump to efficiently remove the muddied soil. In this way, deep coordination of mechanical and hydraulic equipment in terms of time, space and intensity is achieved.

[0136] S5. Closed-loop process for proactive preventive correction:

[0137] Reference Figure 12 When the predicted caisson posture output by the posture evolution prediction model 121 deviates from the preset target posture threshold, the intelligent decision-making system 120 initiates a closed-loop workflow of active preventive correction.

[0138] In the first step, the system calculates the direction and magnitude of the upcoming posture deviation based on the predicted posture and target posture.

[0139] In the second step, based on the calculated deviation direction, the system automatically locks the well wall 14 area on the opposite side as the deviation correction operation area.

[0140] In the third step, the system generates a clear, quantified correction instruction and incorporates it into the final optimal collaborative operation instruction. This correction instruction specifies the activation of the water jet and air curtain systems within the correction operation area, and precisely sets the operation intensity (such as water pressure and air volume) and operation duration based on the deviation value.

[0141] When executing this instruction, the collaborative operation execution system 100 actively and asymmetrically reduces the sinking frictional resistance in a specific area by forming a water or air film between the well wall 14 and the soil. This applies a reverse corrective torque to the caisson before deviation actually occurs, preemptively offsetting predicted posture deviations and achieving proactive, preventative posture control.

[0142] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A construction method for an ultra-large land caisson penetrating an ultra-thick clay layer, characterized in that: The method comprises the following steps: S1. Build an initial digital twin model that includes geological information, caisson structure information, and sensor layout information; S2. During the undrained sinking stage, a multi-dimensional real-time sensing system deployed at the construction site continuously collects data to form a real-time data stream, and uses the real-time data stream to dynamically update the initial digital twin model to obtain a real-time digital twin model; S3, the intelligent decision-making system runs a posture evolution prediction model based on the real-time digital twin model to generate a predicted caisson posture, and runs a ternary balance decision model to perform multi-objective balance calculations based on the predicted caisson posture, and autonomously generates optimal collaborative operation instructions; The operational posture evolution prediction model takes the current caisson posture vector, the planned collaborative action sequence, and the real-time digital twin model as input, and calculates and outputs the predicted caisson posture at the next moment; The ternary balance decision model uses the predicted caisson posture as an important input and solves a comprehensive objective function that includes expected sinking efficiency, comprehensive construction risk, and comprehensive energy consumption to obtain the optimal collaborative operation instructions. The calculation of the comprehensive construction risk factor couples the deviation degree of the predicted caisson posture with the predicted structural stress level. S4, collaborative operation execution system according to the optimal collaborative operation instructions, synchronous control mechanical earth-taking equipment, hydraulic auxiliary equipment and correction equipment to perform collaborative operations; The intelligent decision-making system determines the soil properties of the target excavation area based on the soil mechanical property distribution map in the real-time digital twin model; For areas identified as hard soil, an instruction is generated to prioritize increasing the intensity of high-pressure water jet stirring, using a grab bucket to break up the soil. The suction speed of the mud pump is adjusted in conjunction with the other to match the mud generation rate generated by the coordinated operation of the grab bucket and the high-pressure water gun; S5. If the predicted caisson posture deviates from the preset target, an active preventive correction instruction is incorporated into the generation of the optimal collaborative operation instruction, so as to offset the posture deviation indicated by the predicted caisson posture in advance through collaborative operation.

2. The construction method of a super-large land caisson through an ultra-thick clay layer according to claim 1 is characterized in that: In step S1, the steps of constructing an initial digital twin model including geological information, caisson structure information, and sensor layout information include: The geological information includes soil layer distribution, soil physical and mechanical parameters, and the maximum drainage excavation depth of the clay layer calculated based on historical hydrological data; The caisson structure information is the geometry and structural model of the caisson structure. The caisson structure includes, from bottom to top, a steel shell concrete caisson with wedge-shaped blade feet and a concrete caisson, and is internally provided with a well wall, a partition wall, and a partition wall. The sensor layout information includes the layout position of the mechanical sensor on the grab bucket for obtaining physical and mechanical data, the layout position of the online slurry property analyzer on the mud pipe for obtaining material removal data, and the layout position of the structural stress sensor and the posture sensor in the caisson structure.

3. The construction method of a super-large land caisson through a super-thick clay layer according to claim 1 is characterized in that: In step S2, during the undrained sinking stage, a multi-dimensional real-time perception system deployed at the construction site continuously collects data to form a real-time data stream, and uses the real-time data stream to dynamically update the initial digital twin model. The steps of obtaining the real-time digital twin model include: Using underwater drones equipped with multi-beam sonar to scan the earthwork area to generate geometric data that forms a real-time data stream; Use mechanical sensors to monitor the operating force of the grab bucket in real time to generate physical and mechanical data that constitute a real-time data stream; Mud parameters are monitored using online slurry property analyzers to generate material removal data that constitutes a real-time data stream.

4. The construction method of a super-large land caisson through a super-thick clay layer according to claim 3 is characterized in that: The step of dynamically updating the initial digital twin model using the real-time data stream to obtain the real-time digital twin model also includes: The intelligent decision-making system uses the physical and mechanical data in the real-time data stream as input, inversely generates a soil mechanical property distribution map that characterizes the softness and hardness of the underwater soil, and incorporates the soil mechanical property distribution map into the real-time digital twin model.

5. The construction method of a super-large land caisson penetrating an ultra-thick clay layer according to claim 1 is characterized in that: In step S5, if the predicted caisson posture deviates from the preset target, the steps of incorporating active preventive correction instructions into the generation of the optimal collaborative operation instructions so as to offset the posture deviation indicated by the predicted caisson posture in advance through collaborative operation include: Calculating the predicted attitude deviation direction and deviation amount based on the predicted caisson attitude; Determine the well wall opposite to the deviation direction as the deviation correction operation area; Generate an active preventive correction instruction, which specifies the start-up of the water jet pipe and air curtain pre-buried in the well wall within the correction operation area, and sets the operation intensity of the water jet pipe and air curtain according to the deviation amount, thereby offsetting the posture deviation by reducing the frictional resistance on the opposite side before the deviation occurs.

6. A construction system for an ultra-large land caisson penetrating an ultra-thick clay layer, applied to the method according to any one of claims 1 to 5, characterized in that: The system comprises: Multi-dimensional real-time perception system; Collaborative operation execution system, which includes mechanical earth-moving equipment, hydraulic auxiliary equipment and deviation correction equipment; An intelligent decision-making system is communicatively connected with the multi-dimensional real-time perception system and the collaborative operation execution system, and is used to: Receiving a real-time data stream continuously collected by the multi-dimensional real-time perception system, and dynamically updating the initial digital twin model using the real-time data stream to obtain a real-time digital twin model; Based on the real-time digital twin model, a posture evolution prediction model is run to generate a predicted caisson posture, and a ternary balance decision model is run to perform multi-objective balance calculations based on the predicted caisson posture, and to autonomously generate optimal collaborative operation instructions; If the predicted caisson posture deviates from the preset target, an active preventive correction instruction is incorporated into the generation of the optimal collaborative operation instruction; The optimal collaborative operation instruction is sent to the collaborative operation execution system, so that the mechanical earth-taking equipment, hydraulic auxiliary equipment and correction equipment are synchronously controlled by the collaborative operation execution system to perform collaborative operation.

7. The construction system for an ultra-large land caisson penetrating an ultra-thick clay layer according to claim 6 is characterized in that: The collaborative operation execution system also includes an intelligent soil taking system, and the structure of the intelligent soil taking system is as follows: Large steel distribution beams, Bailey beams, small steel distribution beams and P50 rails are laid in sequence on the shaft wall and the top of the partition wall of the caisson structure, and a gantry crane for lifting the mud pump is set on the P50 rails.

Images