Maritime work wharf prefabricated water outlet and surrounding area foundation settlement cooperative control method

Abstract

The invention discloses a cooperative control method for settlement of a prefabricated drainage port and a surrounding area foundation of a maritime work wharf. The cooperative control method comprises the steps of construction of a hollow pile foundation and a sleeve valve pipe, erection of an intelligent monitoring system, settlement and dynamic load monitoring, starting judgment of grouting deviation correction control, grouting deviation correction control based on artificial intelligence and self-optimization of a model. An intelligent monitoring system and an adjustable pile foundation structure are integrated, a closed-loop control system for monitoring, analysis, decision making and execution is constructed, grouting control parameters are output by adopting a deep neural network model according to settlement and dynamic load data monitored in real time, accurate grouting deviation correction is started and executed, and the grouting deviation correction precision is improved. The traditional passive and static foundation treatment is converted into active and dynamic settlement regulation and control, and the accuracy and reliability of the long-term control effect are ensured.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of foundation treatment, and particularly relates to a method for cooperatively controlling foundation settlement of prefabricated drainage outlets and surrounding areas of a marine wharf. BACKGROUND

[0002] A marine wharf is a key facility in coastal engineering, and its drainage system plays an important role in maintaining the dryness of the wharf surface and ensuring the safety of operations. As an important node of the drainage system, the prefabricated drainage outlet is usually made of large reinforced concrete components. The cooperative settlement control between the prefabricated drainage outlet, the drainage pipeline and the surrounding area foundation directly affects the safety and durability of the overall structure. Under the condition of nearshore soft rock foundation, the prefabricated drainage outlet of the marine wharf mainly faces the following technical problems during the construction and operation stages: 1) The geological conditions in the nearshore area are complex, and soft soil, silt soil and other soft soil layers with low bearing capacity, high compressibility and rheological properties are widely distributed. Such foundations are prone to significant and uneven settlement under load; 2) The prefabricated drainage outlet and the connected drainage pipeline differ in structure, self-weight and stress characteristics, and are prone to non-coordinated deformation on soft foundations, which leads to interface separation, cracking or leakage, and damages the integrity and function of the drainage system; 3) During operation, especially under heavy rain or high-flow drainage conditions, the continuous dynamic impact of high-speed water flow on parts such as the energy dissipation baffle intensifies the disturbance of the soil under and around the foundation, inducing soil particle migration and foundation hollowing, which not only accelerates the uneven settlement of the drainage outlet itself, but also may transmit adverse effects to adjacent pipelines and other structural areas of the wharf, making the settlement coordination problem more complex.

[0003] At present, the conventional treatment methods for the above settlement problems mainly include: 1) using pile foundation to reinforce the key structures such as the drainage outlet to improve the bearing capacity and control the absolute settlement; 2) replacing, preloading or grouting the surrounding soft foundation to reduce post-construction settlement. However, these methods are passive prevention measures, which have obvious limitations: on the one hand, traditional pile foundations lack effective active control and correction ability if the settlement or uneven settlement exceeds the expected value during the operation stage; on the other hand, existing technologies cannot dynamically, accurately and differentially compensate and reinforce the drainage outlet, pipeline and surrounding area foundation during the use of the structure, so as to achieve cooperative settlement control among the three. Especially when local uneven settlement occurs, there is a lack of integrated technical solutions for controllable grouting correction at specific locations and depths.

[0004] Therefore, there is an urgent need in the art for an integrated innovative method that can adapt to the condition of nearshore soft foundation and cooperatively monitor, intelligently analyze and actively control the settlement of prefabricated drainage outlets, connected drainage pipelines and surrounding area foundations of a marine wharf throughout its life cycle, especially during the operation stage, to make up for the shortcomings of existing technologies. SUMMARY

[0005] To achieve the above-mentioned purposes of the invention, the present application provides a method for coordinated control of foundation settlement of prefabricated drainage outlets and surrounding areas of a marine terminal. The core of the method is to build an integrated system that integrates adjustable pile foundation structure and intelligent feedback control, to achieve coordinated control of the prefabricated drainage outlet, the connected drainage pipeline and the foundation settlement of the surrounding area.

[0006] The method for coordinated control of foundation settlement of prefabricated drainage outlets and surrounding areas of a marine terminal comprises the following steps: S1. Construction of hollow pile foundation and sleeve valve pipe Before hoisting the prefabricated drainage outlet structure into place, a number of hollow piles are constructed at a predetermined position below the bottom plate of the prefabricated drainage outlet structure as a foundation; the hollow piles are made of reinforced concrete and are arranged vertically, with the top part being firmly connected to the bottom plate of the prefabricated drainage outlet structure through pre-embedded steel bars or grouting material, forming a foundation structure that has both bearing capacity and active settlement control channel during operation; A sleeve valve pipe is pre-embedded at the central position inside each hollow pile, with one-way valve flaps arranged at a predetermined interval along the length direction of the sleeve valve pipe, and the valve flaps can be opened or closed; a shell material is filled in the annular gap between the outer wall of the sleeve valve pipe and the inner wall of the hollow pile; when grouting is performed, the valve flap at the corresponding position is opened, and the grouting liquid penetrates the shell material and then enters the soil mass at the side or end of the pile through the pre-reserved channel of the hollow pile; S2. Erection of intelligent monitoring system An intelligent monitoring system is constructed, including a data acquisition module, a central processing module, an execution module and a historical database, and information exchange and instruction transmission between modules are performed through a high-speed data bus; the data acquisition module includes a settlement observation instrument and an accelerometer, which are used to acquire real-time settlement and acceleration data respectively, and further calculate a set of monitoring parameters I 1 ; the central processing module is deployed on a local computing device of the terminal, responsible for receiving monitoring data, running a built-in deep neural network model and generating a set of grouting control parameters I 4 ; the set of grouting control parameters I 4 includes the grouting amount h of the first sleeve valve pipe R h1 , grouting pressure R h2 , grouting depth R h3 and expected grouting uplift R act ; the execution module includes an intelligent grouting pump, a flow meter, a pressure sensor and a pipeline valve, which controls the grouting process according to the set of grouting control parameters I 4Precise control of the grouting volume, grouting pressure, grouting depth, and valve opening and closing of each sleeve valve pipe enables precise and differentiated foundation compensation and reinforcement; the historical database is deployed on the dock's local computing equipment for data storage. The input parameters of the deep neural network model include a set of monitoring parameters. I 1 Structural design parameter set I 2 and formation parameter set I 3 The output parameters are the grouting control parameter set. I 4 That is, to establish through training I 1 , I 2 , I 3 and I 4 The nonlinear mapping relationship between them; the set of structural design parameters I 2 Obtained from engineering survey and design data, including the self-weight of the prefabricated drainage outlet structure. A 1 surcharge pressure A 2 Length of base A 3 and bottom width A 4 Length of hollow piles A 5 Pile diameter A 6 and layout spacing A 7 The set of formation parameters I 3 Obtained from engineering survey and design data, including the first k Cohesion of layered rock and soil B k1 internal friction angle B k2 Compression modulus B k3 Dry density B k4 Porosity B k5 Moisture content B k6 Permeability coefficient B k7 Compared to Poisson B k8 ; S3. Settlement and Dynamic Load Monitoring: Settlement monitoring points are set at four corners of the precast drainage outlet structure, two edge points at the interface between the precast drainage outlet structure and the connected drainage pipe, one point at the other end of the connected drainage pipe, and two points on the ground surface on both sides of the precast drainage outlet structure. Settlement data at each settlement monitoring point are continuously monitored using a settlement observation instrument. At the same time, accelerometers are installed on the energy dissipation sill of the precast drainage outlet structure to monitor dynamic load data. The monitoring parameter set I 1 =( S 1 , S 2 , S 3 , S 4 , S 5 , S 6 , S 7 , S 8 , S 9 , S 10 , S 11 The feature vector is a multidimensional feature vector calculated from real-time monitoring data, including: the absolute settlement of the prefabricated drainage outlet structure. S 1 , defined as the monitoring points at the four corners of the prefabricated drainage outlet structure at time . t The average settlement, in millimeters; the uneven settlement of the prefabricated drainage outlet structure itself. S 2 The settlement monitoring points at the four corners of the prefabricated drainage outlet structure are defined as the points at time 10:00 to 11:00. t The difference between the maximum and minimum values, in millimeters; absolute settlement of the interface. S 3 The monitoring point is defined as the monitoring point at the interface between the prefabricated drainage outlet structure and the connected drainage pipe at time 10:00. t The average settlement, in millimeters; the uneven settlement of the connected drainage pipes themselves. S 4 , defined as a settlement monitoring point at one point on the other side of a connected drainage pipe at time . t Settlement and S 3 The difference, in millimeters; interface settlement difference S 5 Defined as S 1 and S 3The difference is expressed in millimeters; the average settlement rate of the prefabricated drainage outlet structure. S 6 Defined as S 1 The rate of change over the past 24 hours, in millimeters per day; the rate of uneven settlement of the prefabricated drainage outlet structure itself. S 7 Defined as S 2 Rate of change in millimeters per day over the past 24 hours; mean surface subsidence S 8 Defined as the settlement monitoring points at two points on the ground surface on both sides of the prefabricated drainage outlet structure at time [time missing]. t The average settlement, in millimeters; the difference in settlement between the prefabricated drainage outlet structure and the surrounding ground surface. S 9 Defined as S 1 and S 8 The difference is expressed in millimeters; the dominant frequency of the dynamic load. S 10 , defined as the frequency component with the strongest energy after Fourier transform of accelerometer monitoring data, is measured in Hertz; dynamic load acceleration amplitude. S 11 Defined as with S 10 The corresponding vibration acceleration amplitude, in m / s² 2 ; S4. Decision for initiating grouting correction control: During the construction or operation of the wharf, the set of monitoring parameters acquired in real time in step S3 I 1 The central processing module determines to initiate grouting correction control when any of the following conditions are met: the interface settlement difference. S 5 Exceeding the preset interface settlement difference safety threshold S 5lim At that time; the uneven settlement of the prefabricated drainage outlet structure itself S 2 Exceeding the preset safety threshold for uneven settlement of the prefabricated drainage outlet structure itself S 2lim At that time; the average settlement rate of the prefabricated drainage outlet structure S 6 Exceeding the preset safety threshold for the average settlement rate of prefabricated drainage outlet structures S 6lim At that time; the uneven settlement rate of the prefabricated drainage outlet structure itself S 7Exceeding the preset safety threshold for uneven settlement rate of the prefabricated drainage outlet structure itself S 7lim hour; in, S 5lim , S 2lim , S 6lim and S 7lim Pre-set based on engineering survey and design data, structural safety specifications, and historical monitoring data; S5. Artificial intelligence-based grouting correction control: After step S4 determines that grouting correction control should be initiated, the central processing module will monitor the parameter set in real time. I 1 Structural design parameter set I 2 and formation parameter set I 3 The input is fed into the trained deep neural network model to calculate and output the grouting control parameter set under the current working condition. I 4 The execution module is based on I 4 The instructions drive the grouting equipment to perform grouting; S6. Model self-optimization: After each grouting correction control operation, the central processing module records the set of grouting control parameters for the actual grouting operation. I 4 and the actual grouting lift obtained through the data acquisition module R act and the corresponding monitoring parameter set I 1 Structural design parameter set I 2 and formation parameter set I 3 These new training samples are collectively added and stored in the historical database. Whenever a preset time period has elapsed or the number of new training samples accumulates to a preset threshold, the central processing module uses the new training samples to incrementally train the deep neural network model. The Adam optimizer is used to iteratively update the network weights and bias parameters, thereby achieving online self-optimization of model parameters and continuous improvement of prediction accuracy.

[0007] Preferably, the deep neural network model adopts a feedforward multilayer perceptron architecture, including an input layer, a hidden layer, an output layer, and a physical constraint layer connected in sequence; The number of nodes in the input layer is consistent with the total dimension of the input parameters, and it is responsible for receiving... I1 , I 2 and I 3 All parameters are processed using the Min-Max normalization method, and all parameters are mapped to the [0,1] interval. The hidden layer consists of at least three fully connected layers, with the number of neurons in each layer decreasing progressively to extract higher-order features. Specifically, the number of neurons in the first fully connected layer is set to 1.5 times the total dimension of the input parameters, and the number of neurons in each subsequent layer decreases by 50%. The hidden layer uses the ReLU function as the activation function to enhance the non-linear expressive power of the model and alleviate the gradient vanishing problem. At the same time, a batch normalization layer is introduced between each hidden layer to improve the stability and convergence speed of the model training. The number of neurons in the output layer and I 4 The dimensions are consistent, used to output the initial grouting control parameter set. I 4 ; The physical constraint layer is connected after the output layer and is used for the initial grouting control parameter set. I 4 Physical rationality verification and constraints were performed, including: grouting volume. R h1 Apply constraints, require R h1 The grouting pressure must not exceed the theoretical maximum volume calculated based on the target uplift, the volume of the affected soil, and the porosity, to prevent over-grouting; R h2 Apply constraints, require R h2 The pressure should be greater than the minimum starting pressure for soil grouting, but less than the maximum allowable pressure required by the soil's splitting strength and structural safety; regarding the grouting depth... R h3 Apply constraints, require R h3 It matches the actual depth range of the sleeve valve tube.

[0008] Preferably, the training process of the deep neural network model is as follows: First, based on the geological survey data and structural design parameters of this project, a refined finite element numerical model is established, and simulations are performed under different grouting control parameter sets. I 4 The foundation and structural response under action generates a large amount of training sample data; each training sample contains I 1 , I 2 , I 3A set of practical and effective grouting control parameters I 4act and the corresponding actual grouting lift R act The initial training sample size should be no less than 1000 sets to ensure coverage of different geological conditions, structural forms, and load conditions. The model training uses the Adam optimization algorithm with mean squared error as the loss function. L mse Perform iterative optimization; loss function L mse As shown in the following formula:

[0009] in, I 4pre The set of grouting control parameters predicted by the model. I 4act A practical and effective set of grouting control parameters; M for I 4pre or I 4act The dimensions are as follows: During training, the Adam optimizer is used for iterative optimization, with the initial learning rate set to 0.001 and the batch size set to 64; During training, when the validation set loss no longer decreases for 10 consecutive epochs, the learning rate is halved; If the validation loss does not improve for 20 consecutive epochs, training is terminated early; After training is completed, the final network weights and bias parameters are saved and deployed to the central processing module.

[0010] In summary, compared with existing technologies, the beneficial effects of this invention are as follows: Addressing the difficulty of implementing dynamic, precise, and differentiated foundation settlement compensation and reinforcement for prefabricated drainage outlets in offshore wharves throughout their entire lifecycle using existing technologies, this invention proposes a collaborative control method for foundation settlement between prefabricated drainage outlets and surrounding areas in offshore wharves. This method includes the construction of hollow pile foundations and sleeve valves, the establishment of an intelligent monitoring system, settlement and dynamic load monitoring, the initiation determination of grouting correction control, artificial intelligence-based grouting correction control, and model self-optimization, achieving the following breakthrough improvements: 1) Realized proactive and intelligent settlement control: By integrating an intelligent monitoring system with an adjustable pile foundation structure, a closed-loop control system for monitoring, analysis, decision-making and execution was constructed. This system can initiate and execute precise grouting correction based on real-time monitored settlement and dynamic load data, transforming traditional passive and static foundation treatment into proactive and dynamic settlement control, significantly improving the ability of prefabricated drainage outlet structures of marine wharves to cope with uncertain settlement development. 2) Improved accuracy and coordination of settlement control: Based on the grouting control parameters output by the deep neural network model, quantitative and pressure-controlled precise grouting can be performed for specific pile foundations and specific depths; This differentiated compensation and reinforcement method can effectively lift the local area where uneven settlement occurs, thereby ensuring the settlement coordination among the prefabricated drainage outlet structure, the connected drainage pipe and the surrounding foundation, and effectively avoiding defects such as interface separation and cracking; 3) Enhanced adaptability under complex load conditions: By monitoring the dynamic load at the drainage outlet energy dissipation sill using accelerometers and using the frequency and amplitude of the dynamic load as input parameters of the neural network model, the control strategy can comprehensively consider the cumulative disturbance effect of cyclic dynamic load on the foundation soil during operation, making settlement control more in line with the actual complex working conditions of marine wharves and improving the applicability and reliability of the method. 4) Possesses adaptive optimization capabilities for soil performance degradation: During the wharf's operation, continuous dynamic loads lead to the deterioration of the physical and mechanical properties of the soil surrounding the drainage outlet foundation, such as changes in effective stress, soil softening, and evolution of pore structure. This time-varying characteristic causes static prediction models based on initial geological parameters to gradually fail. The self-optimization mechanism of this invention can dynamically capture and learn the deterioration trend of soil performance by continuously collecting actual load and settlement response data. Through incremental training, the model continuously corrects its internal mapping relationship, ensuring that the generated grouting control strategy always matches the current actual soil state, thereby effectively combating the foundation performance degradation caused by the cumulative effect of dynamic loads and ensuring the accuracy and reliability of long-term control effects. 5) Ensures the physical rationality and engineering safety of the control strategy: A physical constraint layer is introduced into the deep neural network model to verify the theoretical range and engineering safety of the grouting parameters output by the model. This design effectively prevents irrational outputs that may be generated due to the pure data-driven model, such as over-grouting and pressure exceeding limits. It ensures the feasibility and safety of the control measures in terms of physical principles and engineering specifications, and improves the engineering practical value of the entire system. Attached Figure Description

[0011] Figure 1 This is a flowchart illustrating the method for coordinated control of prefabricated drainage outlets and surrounding ground settlement at marine engineering wharves, as shown in an embodiment of the present invention. Figure 2 This is a schematic diagram of the prefabricated drainage outlet structure of the marine wharf as shown in an embodiment of the present invention; Figure 3 This is a cross-sectional schematic diagram of the hollow pile foundation and sleeve valve pipe shown in an embodiment of the present invention; Figure 4 This is a connection diagram of the intelligent monitoring system shown in an embodiment of the present invention; Among them, 1-prefabricated drainage outlet structure, 11-energy dissipation sill, 12-bottom plate, 21-hollow pile, 22-sleeve valve pipe, 23-casing material, 24-valve disc; 3-data acquisition module, 31-settlement observation instrument, 32-accelerometer, 4-central processing module, 5-execution module, 51-intelligent grouting pump, 52-flow meter, 53-pressure sensor, 54-pipeline valve, 6-historical database. Detailed Implementation

[0012] The specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments given herein are for illustration and explanation only and are not intended to limit the present invention.

[0013] This application discloses, as follows: Figures 1-4 The method for coordinated control of prefabricated drainage outlets and surrounding ground settlement at offshore wharves, as shown, includes the following steps: S1. Construction of hollow pile foundation and sleeve valve pipe: Before the prefabricated drainage outlet structure 1 is hoisted into place, several hollow piles 21 are constructed as a foundation at a preset position below its base plate 12; the hollow piles 21 are made of reinforced concrete, are set vertically, and are firmly connected to the base plate 12 of the prefabricated drainage outlet structure by pre-embedded steel bars or grouting material at the top, forming a foundation structure that has both load-bearing capacity and active settlement control channel during operation.

[0014] A sleeve valve pipe 22 is pre-embedded at the center of each hollow pile 21. One-way valve flaps 24 are arranged at preset intervals along the length of the sleeve valve pipe 22. A low-strength, easily crackable casing material 23 is filled in the annular gap between the outer wall of the sleeve valve pipe 22 and the inner wall of the hollow pile 21. The casing material 23 is preferably a mixture of cement and bentonite. Under normal conditions, the casing material 23 remains closed. During high-pressure grouting, the valve flaps 24 at the corresponding positions are opened, thus forming a multi-stage grouting channel that can be opened and closed as needed. After the grout penetrates the casing material 23, it is pressed into the soil on the pile side or pile end through the reserved channel of the hollow pile 21, achieving precise grouting correction. After grouting is completed, the casing material 23 automatically fills the grouting channel under gravity.

[0015] S2. Construction of an intelligent monitoring system: An intelligent monitoring system is constructed, including a data acquisition module 3, a central processing module 4, an execution module 5, and a historical database 6. Information exchange and command transmission between modules are achieved via a high-speed data bus. The data acquisition module 3 includes a settlement observation instrument 31 and an accelerometer 32, used to acquire settlement and acceleration data in real time, respectively, and further calculate the monitoring parameter set. I 1 The central processing module 4 is deployed on the local computing equipment at the dock, and is responsible for receiving monitoring data, running the built-in deep neural network model, and generating a set of grouting control parameters. I4 The grouting control parameter set I 4 Including the h Grouting volume of root sleeve valve pipe R h1 Grouting pressure R h2 Grouting depth R h3 and the expected grouting lift R act The execution module 5 includes an intelligent grouting pump 51, a flow meter 52, a pressure sensor 53, and pipeline valves 54, according to... I 4 Precise control of the grouting volume, grouting pressure, and grouting depth of each sleeve valve pipe enables precise and differentiated foundation compensation and reinforcement; the historical database 6 is deployed on the local computing equipment at the dock for data storage.

[0016] The input parameters of the deep neural network model include a set of monitoring parameters. I 1 Structural design parameter set I 2 and formation parameter set I 3 The output parameters are the grouting control parameter set. I 4 That is, to establish through training I 1 , I 2 , I 3 and I 4 The nonlinear mapping relationship between them; the set of structural design parameters I 2 Obtained from engineering survey and design data, including the self-weight of the prefabricated drainage outlet structure. A 1 surcharge pressure A 2 Length of base A 3 and bottom width A 4 Length of hollow piles A 5 Pile diameter A 6 and layout spacing A 7 The set of formation parameters I 3 Obtained from engineering survey and design data, including the first k Cohesion of layered rock and soil B k1internal friction angle B k2 Compression modulus B k3 Dry density B k4 Porosity B k5 Moisture content B k6 Permeability coefficient B k7 Compared to Poisson B k8 .

[0017] In specific implementation, the deep neural network model adopts a feedforward multilayer perceptron architecture, which includes an input layer, a hidden layer, an output layer, and a physical constraint layer connected in sequence.

[0018] The number of nodes in the input layer is consistent with the total dimension of the input parameters, and it is responsible for receiving... I 1 , I 2 and I 3 All parameters are processed using the Min-Max normalization method, and all parameters are mapped to the [0,1] interval.

[0019] The hidden layer consists of at least three fully connected layers, with the number of neurons in each layer decreasing progressively to extract higher-order features. Specifically, the number of neurons in the first fully connected layer is set to 1.5 times the total dimension of the input parameters, and the number of neurons in each subsequent layer decreases by 50%. The hidden layer uses the ReLU function as the activation function to enhance the non-linear expressive power of the model and alleviate the gradient vanishing problem. At the same time, a batch normalization layer is introduced between each hidden layer to improve the stability and convergence speed of the model training.

[0020] The number of neurons in the output layer and I 4 The dimensions are consistent, used to output the initial grouting control parameter set. I 4 .

[0021] The physical constraint layer is connected after the output layer and is used for the initial grouting control parameter set. I 4 Physical rationality verification and constraints were performed, including: grouting volume. R h1 Apply constraints, require R h1 The grouting pressure must not exceed the theoretical maximum volume calculated based on the target uplift, the volume of the affected soil, and the porosity, to prevent over-grouting; R h2Apply constraints, require R h2 The pressure should be greater than the minimum starting pressure for soil grouting, but less than the maximum allowable pressure required by the soil's splitting strength and structural safety; regarding the grouting depth... R h3 Apply constraints, require R h3 It matches the actual depth range of the sleeve valve tube.

[0022] In specific implementation, the training process of the deep neural network model is as follows: First, based on the geological survey data and structural design parameters of this project, a refined finite element numerical model is established, and simulations are performed under different grouting control parameter sets. I 4 The foundation and structural response under action generates a large amount of training sample data; each training sample contains I 1 , I 2 , I 3 A set of practical and effective grouting control parameters I 4act and the corresponding actual grouting lift R act The initial training sample size should be no less than 1000 sets to ensure coverage of different geological conditions, structural forms, and load conditions. The model training uses the Adam optimization algorithm with mean squared error as the loss function. L mse Perform iterative optimization; loss function L mse As shown in the following formula: (1) in, I 4pre The set of grouting control parameters predicted by the model. I 4act A practical and effective set of grouting control parameters; M for I 4pre or I 4act The dimensions are as follows: During training, the Adam optimizer is used for iterative optimization, with the initial learning rate set to 0.001 and the batch size set to 64; During training, when the validation set loss no longer decreases for 10 consecutive epochs, the learning rate is halved; If the validation loss does not improve for 20 consecutive epochs, training is terminated early; After training is completed, the final network weights and bias parameters are saved and deployed to the central processing module.

[0023] S3. Settlement and dynamic load monitoring: Settlement monitoring points are set at the four corners of the precast drainage outlet structure, the two edge points at the interface between the precast drainage outlet structure and the connected drainage pipe, the one point at the other end of the connected drainage pipe, and the two points on the ground surface on both sides of the precast drainage outlet structure. The settlement data of each settlement monitoring point is continuously monitored by the settlement observation instrument 31. At the same time, an accelerometer 32 is installed on the energy dissipation sill 11 of the precast drainage outlet structure to monitor the dynamic load data.

[0024] The monitoring parameter set I 1 =( S 1 , S 2 , S 3 , S 4 , S 5 , S 6 , S 7 , S 8 , S 9 , S 10 , S 11 The feature vector is a multidimensional feature vector calculated from real-time monitoring data, including: the absolute settlement of the prefabricated drainage outlet structure. S 1 , defined as the monitoring points at the four corners of the prefabricated drainage outlet structure at time . t The average settlement, in millimeters; the uneven settlement of the prefabricated drainage outlet structure itself. S 2 The settlement monitoring points at the four corners of the prefabricated drainage outlet structure are defined as the points at time 10:00 to 11:00. t The difference between the maximum and minimum values, in millimeters; absolute settlement of the interface. S 3 The monitoring point is defined as the monitoring point at the interface between the prefabricated drainage outlet structure and the connected drainage pipe at time 10:00. t The average settlement, in millimeters; the uneven settlement of the connected drainage pipes themselves. S 4 , defined as a settlement monitoring point at one point on the other side of a connected drainage pipe at time . t Settlement and S 3 The difference, in millimeters; interface settlement difference S 5 Defined as S 1 andS 3 The difference is expressed in millimeters; the average settlement rate of the prefabricated drainage outlet structure. S 6 Defined as S 1 The rate of change over the past 24 hours, in millimeters per day; the rate of uneven settlement of the prefabricated drainage outlet structure itself. S 7 Defined as S 2 Rate of change in millimeters per day over the past 24 hours; mean surface subsidence S 8 Defined as the settlement monitoring points at two points on the ground surface on both sides of the prefabricated drainage outlet structure at time [time missing]. t The average settlement, in millimeters; the difference in settlement between the prefabricated drainage outlet structure and the surrounding ground surface. S 9 Defined as S 1 and S 8 The difference is expressed in millimeters; the dominant frequency of the dynamic load. S 10 , defined as the frequency component with the strongest energy after Fourier transform of accelerometer monitoring data, is measured in Hertz; dynamic load acceleration amplitude. S 11 Defined as with S 10 The corresponding vibration acceleration amplitude, in m / s² 2 .

[0025] S4. Criteria for initiating grouting correction control: During wharf construction or operation, the set of monitoring parameters acquired in real time in step S3 is used as the starting point. I 1 The central processing module determines to initiate grouting correction control when any of the following conditions are met: the interface settlement difference. S 5 Exceeding the preset interface settlement difference safety threshold S 5lim At that time; the uneven settlement of the prefabricated drainage outlet structure itself S 2 Exceeding the preset safety threshold for uneven settlement of the prefabricated drainage outlet structure itself S 2lim At that time; the average settlement rate of the prefabricated drainage outlet structure S 6 Exceeding the preset safety threshold for the average settlement rate of prefabricated drainage outlet structures S 6lim At that time; the uneven settlement rate of the prefabricated drainage outlet structure itself S7 Exceeding the preset safety threshold for uneven settlement rate of the prefabricated drainage outlet structure itself S 7lim At that time; among them, S 5lim , S 2lim , S 6lim and S 7lim Based on pre-set engineering survey and design data, structural safety specifications, and historical monitoring data, in specific implementation, [the following is taken / taken]. S 5lim =15mm, S 2lim =10mm, S 6lim =5mm / d, S 7lim =2mm / d.

[0026] S5. Artificial Intelligence-Based Grouting Correction Control: After step S4 determines that grouting correction control should be initiated, the central processing module will monitor the parameter set in real time. I 1 Structural design parameter set I 2 and formation parameter set I 3 The input is fed into the trained deep neural network model to calculate and output the grouting control parameter set under the current working condition. I 4 The execution module is based on I 4 The instructions drive the grouting equipment to perform precise grouting; specifically, the grouting fluid is injected at the grouting pressure. R h2 Lower penetration grouting depth R h3 The casing material is injected into the soil at the pile side or pile tip in a controlled manner, forming a sheet-like or vein-like solidified body in the soil, and generating a squeezing and expansion effect on the surrounding soil, thereby generating an upward lifting force at the pile-soil interface.

[0027] S6. Model self-optimization: After each grouting correction control operation, the central processing module records the set of grouting control parameters for the actual grouting operation. I 4 and the actual grouting lift obtained through the data acquisition module R act and the corresponding monitoring parameter set I 1 Structural design parameter set I 2 and formation parameter set I3 These new training samples are collectively stored in the historical database. Every three months or when the number of new training samples reaches a preset threshold of 10, the central processing module uses the new training samples to incrementally train the deep neural network model. The Adam optimizer is used to iteratively update the network weights and bias parameters, thereby achieving online self-optimization of model parameters and continuous improvement of prediction accuracy.

[0028] The above describes one or more embodiments of the present invention in a relatively specific and detailed manner, but it should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of this patent should be determined by the appended claims.

Claims

1. A method for co-control of foundation settlement of a prefabricated drainage outlet of a marine wharf and a surrounding area, characterized in that, Comprising the following steps: S1, construction of hollow pile foundation and sleeve valve pipe: a plurality of hollow piles are constructed below the bottom plate of the prefabricated drainage port structure; the hollow piles are vertically arranged and the top is firmly connected with the bottom plate of the prefabricated drainage port structure; a sleeve valve pipe is embedded in the central position inside each hollow pile, the sleeve valve pipe is provided with one-way openable and closable valve flap along the length direction; the annular gap between the outer wall of the sleeve valve pipe and the inner wall of the hollow pile is filled with a shell material; S2, erection of intelligent monitoring system: an intelligent monitoring system is constructed, including a data acquisition module, a central processing module, an execution module and a historical database; the data acquisition module includes a settlement observation instrument and an accelerometer, which are respectively used for acquiring real-time settlement and acceleration data; the central processing module is deployed on a local computing device of the wharf, and is responsible for receiving monitoring data, running a built-in deep neural network model and generating a grouting control parameter set I 4 ; the execution module includes an intelligent grouting pump, a flow meter, a pressure sensor and a pipeline valve; the historical database is deployed on the local computing device of the wharf, and is used for storing data; S3, settlement and dynamic load monitoring: settlement monitoring points are arranged around the prefabricated drainage port structure and the surrounding area, and the settlement data of each settlement monitoring point is continuously monitored through a settlement observation instrument; at the same time, an accelerometer is installed on the energy dissipation ridge of the prefabricated drainage port structure to monitor the dynamic load data; S4, start-up determination of grouting deviation correction control: during wharf construction or operation, for the monitoring parameter set acquired in real time in step S3 I 1 When the preset condition is met, the central processing module determines to start the grouting deviation correction control. S5, grouting deviation control based on artificial intelligence: after step S4 determines to start the grouting deviation control, the central processing module will monitor the parameter set in real time I 1 , the structural design parameter set I 2 and the formation parameter set I 3 Input into the trained deep neural network model, calculate and output the current working condition I 4 ; the execution module drives the grouting equipment to grout according to the instruction of I 4 ; S6, self-optimization of the model: after each execution of the grouting deviation control, the central processing module records the actual execution grouting operation of the model I 4 and the actual grouting lifting amount R act , and the corresponding I 1 , I 2 and I 3 together constitute a new training sample, and the deep neural network model is incrementally trained.

2. The method according to claim 1, wherein, The input parameters of the deep neural network model include a monitoring parameter set I 1 , a structure design parameter set I 2 , and a formation parameter set I 3 , and the output parameters are a grouting control parameter set I 4 , that is, a nonlinear mapping relationship between I 1 , I 2 , I 3 and I 4 is established through training. The set of grouting control parameters I 4 The grouting control parameter set h The grouting amount of the root sleeve valve pipe R h1 The grouting pressure R h2 The grouting depth R h3 And the expected grouting uplift amount R act The deep neural network model adopts a feedforward multilayer perception architecture, including an input layer, a hidden layer, an output layer and a physical constraint layer connected in turn; The number of nodes of the input layer is consistent with the total dimension of input parameters, responsible for receiving I 1 , I 2 and I 3 all parameters, and adopting Min-Max normalization processing method to map all parameters to [0, 1] interval; The hidden layer is provided with not less than three full connection layers, and the number of neurons of each layer is gradually reduced to gradually extract high-order features; specifically, the number of neurons of the first full connection layer is set to 1.5 times the total dimension of the input parameters, and the number of neurons of each subsequent layer is reduced by 50%; the ReLU function is used as the activation function of the hidden layer to enhance the nonlinear expression ability of the model and relieve the gradient vanishing problem; at the same time, a batch normalization layer is introduced between each hidden layer to improve the stability and convergence speed of the model training; The number of neurons of the output layer is consistent with the dimension of the input layer, and is used for outputting a preliminary grouting control parameter set I 4 ; and I 4 ; and The physical constraint layer is connected after the output layer, for controlling the preliminary grouting control parameter set I 4 Physical rationality verification and constraint are carried out, including: grouting quantity R h1 Constraint is applied, requiring R h1 Theoretical maximum volume calculated based on target lifting amount, affected soil volume and porosity, to prevent excessive grouting; grouting pressure R h2 Constraint is applied, requiring R h2 Greater than the minimum starting pressure of soil grouting, while less than the maximum allowable pressure required by soil anti-splitting strength and structural safety; grouting depth R h3 Constraint is applied, requiring R h3 Matching with the actual depth range of the sleeve valve pipe.

3. The method according to any one of claims 1-2, wherein, The training process of the deep neural network model is as follows: firstly, according to the geological survey data and structural design parameters of the project, a refined finite element numerical model is established, and a large amount of training sample data is generated by simulating the response of the foundation and structure under different sets of grouting control parameters I 4 ​ Each training sample contains I 1 , I 2 , I 3 , the actual effective grouting control parameter set I 4act , and the corresponding actual grouting uplift R act ; the total number of initial training samples is not less than 1000 groups to ensure covering different geological conditions, structural forms and load working conditions; the model training adopts Adam optimization algorithm, and the mean square error is taken as the loss function L mse for iterative optimization; the loss function L mse is as follows: （1） wherein, I 4pre a set of grouting control parameters predicted by the model, I 4act a set of grouting control parameters actually effective; M for I 4pre or I 4act dimension; Adam optimizer is used for iterative optimization in training, the initial learning rate is set to 0.001, and the batch size is set to 64; during the training process, when the validation set loss no longer decreases for 10 consecutive epochs, the learning rate is halved; if the validation loss does not improve for 20 consecutive epochs, the training is terminated in advance; after the training is completed, the final network weight and bias parameters are saved and deployed to the central processing module.

4. The method for controlling the foundation settlement of the prefabricated drainage outlet and the surrounding area of the marine terminal according to claim 1, characterized in that, said set of structure design parameters I 2 obtained from engineering surveying and designing data, including the self-weight of the prefabricated drainage port structure A 1 , the surcharge pressure A 2 , the bottom surface length A 3 , and the bottom surface width A 4 , the pile length of the hollow pile A 5 , the pile diameter A 6 , and the layout spacing A 7 ; said set of stratum parameters I 3 obtained from engineering surveying and designing data, including the cohesion of the first k stratum B k1 , the internal friction angle B k2 , the compression modulus B k3 , the dry density B k4 , the porosity B k5 , the water content B k6 , the permeability coefficient B k7 , and the Poisson's ratio B k8 .

5. The method for the coordinated control of the foundation settlement of the prefabricated drainage outlet and the surrounding area of the maritime wharf according to claim 1, characterized in that, The monitoring parameter set I 1 =( S 1 , S 2 , S 3 , S 4 , S 5 , S 6 , S 7 , S 8 , S 9 , S 10 , S 11 ) is a multi-dimensional feature vector calculated from real-time monitoring data, including: absolute settlement of prefabricated drainage port structure S 1 , defined as the average value of the settlement of the monitoring points of the four corner points of the prefabricated drainage port structure at time t , with the unit of millimeters; self uneven settlement of prefabricated drainage port structure S 2 , defined as the difference between the maximum and minimum values of the settlement monitoring points of the four corner points of the prefabricated drainage port structure at time t , with the unit of millimeters; interface absolute settlement S 3 , defined as the average value of the settlement of the monitoring points of the two edge points at the interface between the prefabricated drainage port structure and the connected drainage pipeline at time t , with the unit of millimeters; self uneven settlement of connected drainage pipeline S 4 , defined as the difference between the settlement of the settlement monitoring point of one point at the other end of the connected drainage pipeline and S 3 at time t , with the unit of millimeters; interface settlement difference S 5 , defined as the difference between S 1 and S 3 , with the unit of millimeters; average settlement rate of prefabricated drainage port structure S 6 , defined as the change rate of S 1 in the last 24 hours, with the unit of millimeters per day; self uneven settlement rate of prefabricated drainage port structure S 7 , defined as S 2 Rate of change in millimeters per day over the past 24 hours; mean surface subsidence S 8 Defined as the settlement monitoring points at two points on the ground surface on both sides of the prefabricated drainage outlet structure at time [time missing]. t The average settlement, in millimeters; the difference in settlement between the prefabricated drainage outlet structure and the surrounding ground surface. S 9 Defined as S 1 and S 8 The difference is expressed in millimeters; the dominant frequency of the dynamic load. S 10 , defined as the frequency component with the strongest energy after Fourier transform of accelerometer monitoring data, is measured in Hertz; dynamic load acceleration amplitude. S 11 Defined as with S 10 The corresponding vibration acceleration amplitude, in m / s² 2 .

6. The method for the coordinated control of the foundation settlement of the prefabricated drainage outlet and the surrounding area of the maritime wharf according to claim 1, characterized in that, In step S3, settlement monitoring points are arranged at four corner points of the prefabricated drainage port structure, two edge points at the interface between the prefabricated drainage port structure and the connected drainage pipeline, one point at the other end of the connected drainage pipeline, and two points on the ground surface on both sides of the prefabricated drainage port structure.

7. The method according to any one of claims 1 to 6, wherein, The preset condition in step S4 is satisfied when any of the following conditions is satisfied: the interface settlement difference S 5 exceeds a preset interface settlement difference safety threshold S 5lim ; the self-uniform settlement amount of the prefabricated drainage port structure S 2 exceeds a preset self-uniform settlement amount safety threshold of the prefabricated drainage port structure S 2lim ; the average settlement rate of the prefabricated drainage port structure S 6 exceeds a preset average settlement rate safety threshold of the prefabricated drainage port structure S 6lim ; the self-uniform settlement rate of the prefabricated drainage port structure S 7 exceeds a preset self-uniform settlement rate safety threshold of the prefabricated drainage port structure S 7lim ; wherein, S 5lim , S 2lim , S 6lim and S 7lim Based on the engineering survey design data, structure safety specification and historical monitoring data.

Images