A method for controlling sinking of a steel pipe pile of a container wharf in a deep-water port area

Abstract

The application relates to the technical field of ocean engineering and automatic control, and discloses a deepwater port container wharf steel pipe pile sinking control method, which comprises the following steps: deploying a multi-source sensor network to construct an informationized pile sinking operation system and constructing the informationized pile sinking operation system, collecting and fusing multi-source sensing data to generate a unified optimal state vector, online identifying and updating a dynamic model to form an adaptive model reflecting the current real dynamic characteristics of the system, making a prospective multi-objective optimization decision based on the adaptive model to obtain an optimal control instruction, executing the instruction in multiple scales and controlling through closed-loop feedback iteration. Through the construction of the multi-source sensor network and the adaptive dynamic model, the application realizes the accurate control of the attitude deviation in the steel pipe pile sinking process under the complex marine environment, the collaborative optimization of the energy consumption and the safety constraint, and overcomes the problems of insufficient precision and low efficiency caused by model mismatch in the traditional control method.

Description

Technical Field

[0001] This invention relates to the field of marine engineering and automation control technology, specifically to a method for controlling the driving of steel pipe piles in a container terminal in a deep-water port area. Background Technology

[0002] The steel pipe pile driving operation at deep-water container terminals is a critical engineering step in constructing the foundation support structure of the terminal by driving steel pipe piles into the ground through specific processes in port areas with large water depths. The quality of the operation directly affects the terminal's load-bearing capacity, wind and wave resistance, and long-term service life. Especially in complex marine environments such as strong winds, giant waves, and rapid currents, the requirements for attitude control, stress monitoring, and dynamic adjustment during the pile driving process are extremely high.

[0003] Existing technologies typically employ open-loop control or simple closed-loop feedback control based on empirical parameters when dealing with such operations. Specifically, this involves deploying a single type of sensor, such as an inclinometer, to monitor the pile's attitude, combining this with a pre-set pile driving dynamics model to calculate control parameters, and relying on operators to manually adjust the hammering energy of construction equipment, such as a hydraulic hammer, based on real-time data. This control method depends on the accuracy of the initial model and the operator's experience, and can meet certain pile driving requirements under normal working conditions.

[0004] However, the marine environment in deep-water port areas has significant time-varying and nonlinear characteristics. Factors such as the dynamic coupling of wind, wave, and current loads and the real-time changes in pile-soil interaction make it impossible for existing technologies to accurately capture the dynamic characteristics of the system in real time. The matching degree between the model and the actual working conditions gradually decreases. This defect causes several problems: First, the pile driving posture deviation accumulates as the operation progresses, which can cause the steel pipe pile to tilt beyond the limit, affecting the stability of the wharf structure. Second, it is impossible to balance energy consumption and safety constraints, which can lead to pile damage or energy waste due to excessive hammering, reducing construction efficiency and increasing project costs. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a method for controlling the driving of steel pipe piles in container terminals in deep-water ports, aiming to solve the problems of excessive cumulative deviation of driving posture in deep-water environments and the difficulty in balancing energy consumption and safety constraints.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a method for controlling the driving of steel pipe piles in a deep-water port container terminal, comprising the following steps:

[0007] S1. Deploy a multi-source sensor network for the initial pile driving platform to build an information-based pile driving system capable of digital sensing.

[0008] S2. Based on the information-based pile driving operation system, multi-source sensor data during the pile driving process are collected and fused in real time to generate a unified optimal state vector that characterizes the dynamic characteristics of the system at the current moment.

[0009] S3. Based on the unified optimal state vector, the preset coupled system dynamics model is identified and its parameters are updated online to obtain an adaptive dynamics model that can accurately reflect the current real dynamic characteristics of the system.

[0010] S4. Apply an adaptive dynamics model to perform forward-looking simulations and multi-objective optimization calculations to determine the unique optimal strategic control command that takes into account attitude, energy consumption, and safety.

[0011] S5. Perform multi-scale collaborative analysis and physical execution of the optimal strategic control command, and initiate the next round of control iteration through a closed-loop feedback mechanism.

[0012] Preferably, the step S1 of deploying the multi-source sensor network for the initial pile driving platform includes:

[0013] An inertial measurement unit and a differential global positioning system are installed on the top of the steel pipe pile, and multiple tilt sensors and multiple strain gauges are arranged along the length of the pile.

[0014] A second set of inertial measurement units and a second set of differential global positioning systems are installed on the construction vessel to monitor its own position and orientation, and an environmental monitoring device is integrated to acquire environmental load data of wind, waves, and currents.

[0015] Preferably, the step of fusing the collected multi-source sensor data in step S2 is as follows:

[0016] Based on the system state estimate and state covariance obtained from the previous control cycle, a set of sigma sampling points with deterministic weights is generated.

[0017] The sigma sampling points are substituted into the preset nonlinear system state equation and observation equation respectively for propagation, so as to calculate the prior state estimate and predicted observation value at the current time.

[0018] The Kalman gain is calculated by combining real-time acquired multi-source sensor data with predicted observations.

[0019] Kalman gain is used to correct the prior state estimate in order to generate a unified optimal state vector.

[0020] Preferably, the calculation of the Kalman gain to correct the prior state estimate is achieved by the following formula:

[0021] ;

[0022] In the formula, To unify the optimal state vector, For prior state estimation, For Kalman gain, For real-time acquisition of multi-source sensor data, To predict the observed values.

[0023] Preferably, the step of online identification and parameter update of the preset coupled system dynamics model in step S3 is as follows:

[0024] Based on the unified optimal state vector, the actual output of the system at the current moment is extracted and a regression vector containing the system's historical states and inputs is constructed;

[0025] Calculate the gain matrix at the current time step based on the parameter covariance matrix and regression vector from the previous time step;

[0026] The key parameter vector from the previous time step is corrected using the gain matrix and the actual output of the system at the current time step to obtain the updated key parameter vector at the current time step. This updated key parameter vector is then applied to a preset coupled system dynamics model to obtain an adaptive dynamics model.

[0027] Preferably, the calculation for correcting the key parameter vector from the previous time step is achieved through the following formula:

[0028] ;

[0029] In the formula, This is the updated key parameter vector at the current moment. This is the key parameter vector from the previous time step. Here is the gain matrix. This represents the system's actual output at the current moment. This is the regression vector.

[0030] Preferably, the step S4, which involves applying an adaptive dynamics model for forward-looking deduction and multi-objective optimization calculation, is as follows:

[0031] Based on the adaptive dynamics model, a set of candidate control sequences covering the preset control time domain is generated and the future system state trajectory corresponding to each candidate control sequence is deduced.

[0032] A comprehensive performance evaluation function that integrates attitude deviation, energy consumption index and safety constraints is constructed, and this function is used to quantitatively evaluate each future system state trajectory to obtain its performance evaluation value.

[0033] The candidate control sequence with the best performance evaluation value is selected, and its initial control command is taken as the unique optimal strategic control command.

[0034] Preferably, the comprehensive performance evaluation function is calculated using the following formula:

[0035] ;

[0036] In the formula, This is a comprehensive performance evaluation value. To predict discrete time steps in the time domain, To predict the length of the time domain, For the predicted future system state, For the corresponding expected state, For candidate control inputs, The state weight matrix is... To control the input weight matrix, The penalty function for safety constraints.

[0037] Preferably, the step S5, which involves multi-scale collaborative analysis and physical execution of the optimal strategic control command, comprises:

[0038] The single optimal strategic control command is decomposed into multiple sub-commands that act on different time and space scales;

[0039] Based on multiple sub-instructions, the underlying drive control signals for driving the physical actuator are calculated and sent out to coordinate the completion of the pile driving operation.

[0040] Preferably, the calculation of the underlying drive control signal is achieved through the following control algorithm:

[0041] ;

[0042] In the formula, These are the underlying drive control signals. This is the deviation between the target value of the sub-instruction and the feedback value of the physical actuator. For proportional gain, For integral gain, This is the differential gain.

[0043] This invention provides a method for controlling the driving of steel pipe piles in deep-water container terminals. It has the following beneficial effects:

[0044] 1. This invention achieves comprehensive and accurate perception of the key states of the pile driving system by deploying a multi-source sensor network on the pile body and the construction vessel and fusing them to generate a unified optimal state vector. Combined with the preset safety boundary constraints in the optimization decision-making, it ensures that the control strategy is always executed within the safety domain, which helps to improve the accuracy and structural safety of steel pipe pile driving operations in deep water complex environments.

[0045] 2. This invention uses real-time state vectors to perform online identification and parameter updates of the coupled system dynamic model, constructing an adaptive dynamic model that accurately reflects the current working conditions. This model can dynamically adapt to the real-time changes in pile-soil interaction and environmental loads such as wind, waves, and currents, solving the problem of decreased accuracy caused by model mismatch in traditional control methods and ensuring the robustness of this control method under time-varying and nonlinear working conditions.

[0046] 3. This invention applies a model predictive control framework based on an adaptive model and constructs a comprehensive performance evaluation function that integrates attitude deviation, energy consumption, and safety constraints. This enables forward-looking multi-objective optimization of the control process. It not only pursues single attitude accuracy but also finds the globally optimal strategic control command that takes into account both energy consumption and equipment smoothness, which contributes to the overall economic benefits of pile driving operations.

[0047] 4. This invention constructs a complete closed-loop control architecture of "perception-estimation-identification-decision-execution" and decomposes and issues macro-optimal strategic instructions, thereby achieving highly automated collaborative control. This rolling optimization and closed-loop feedback mechanism enables the system to continuously self-correct and iteratively seek optimization, reducing the dependence on operator experience and improving the automation and intelligence level of the pile driving control method. Attached Figure Description

[0048] Figure 1 This is a schematic diagram of the method flow of the present invention;

[0049] Figure 2 This is a schematic diagram of the physical deployment of the intelligent pile driving operation system of the present invention;

[0050] Figure 3 This is a schematic diagram illustrating the core principles of Model Predictive Control (MPC) and Rolling Optimization of the present invention.

[0051] Figure 4 This is a diagram of the closed-loop control and information flow architecture of the present invention. Detailed Implementation

[0052] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0053] Please see the appendix Figure 1 -Appendix Figure 4 This invention provides a method for controlling the driving of steel pipe piles in a deep-water port container terminal, comprising the following steps:

[0054] S1. Deploy a multi-source sensor network for the initial pile driving platform to build an information-based pile driving system capable of digital sensing.

[0055] Specifically, an inertial measurement unit (IMU) and a high-precision differential global positioning system (DGPS) are preferably installed at the top of the steel pipe pile to accurately determine the spatial attitude parameters and displacement trajectory of the steel pipe pile during the pile driving process.

[0056] The IMU has the ability to acquire pile acceleration and angular velocity data in real time, while the DGPS is used for high-precision positioning compensation. The combination of the two can output the complete six-degree-of-freedom motion information of the pile.

[0057] Furthermore, in order to obtain the deformation response of the pile structure during the pile driving process, it is preferable to arrange several sets of tilt sensors and strain gauges along the axial direction of the steel pipe pile.

[0058] Inclination sensors can detect the bending trend and angle shift of the pile in real time, while strain gauges can obtain the stress and strain state of the pile under longitudinal and lateral forces. These data will serve as an important input basis for constructing a dynamic model of the coupled system.

[0059] On the construction vessel platform, there is also a second set of inertial measurement units and a second set of differential global positioning systems for monitoring the spatial motion of the vessel itself.

[0060] The system configuration ensures that the impact of hull disturbance on the pile driving process under dynamic marine environmental conditions such as waves, wind loads, and currents can be fully identified and fed back to the control system in real time, so as to realize the modeling and compensation of the impact of platform movement on operational accuracy.

[0061] In order to fully capture the coupling effect of environmental factors on the pile driving system, an environmental monitoring device is integrated on the construction vessel. This device includes a sensing submodule for acquiring marine meteorological parameters such as wind speed, wind direction, wave height, and current velocity.

[0062] The aforementioned environmental load data, along with the structural response and platform attitude data, will be input into the subsequent state fusion and dynamic model update process.

[0063] Multi-source sensors are connected to form an integrated sensor network via high-bandwidth wired or wireless communication links, and then connected to an embedded edge computing platform or a remote cloud data server to achieve synchronous data acquisition, preliminary preprocessing and timestamp calibration.

[0064] To ensure the quality and timeliness of the acquired data, the system preferably integrates a time synchronization module to coordinate the alignment of various sensor data frames in the time domain and avoid estimation deviations caused by asynchronous acquisition.

[0065] In the initial stage of system operation, a sensor calibration model will be constructed based on the preset installation parameters and deployment points. An initial data set will be obtained through trial operation, which will be used to identify and compensate for the measurement errors of each sensor, effectively enhancing the overall observation accuracy and status recognition capability of the system.

[0066] The aforementioned multi-source sensing information will be transmitted as an input stream to the state estimation module of the information-based pile driving operation system, serving as the core data foundation for subsequent state vector construction, dynamic model updating, and control command generation.

[0067] This implementation method deploys a multi-source sensor network on the pile driving platform to establish a comprehensive information acquisition channel connecting the structure, platform, and environment. This provides the necessary data support and environmental modeling foundation for real-time state estimation, online system identification, and the generation of optimal control strategies.

[0068] S2. Based on the information-based pile driving operation system, multi-source sensor data during the pile driving process are collected and fused in real time to generate a unified optimal state vector that characterizes the dynamic characteristics of the system at the current moment.

[0069] Specifically, firstly, multi-source sensor data includes, but is not limited to, the following:

[0070] Pile driving displacement data: acquired through displacement sensors installed on the working platform or pile driving equipment, including pile driving depth, axial displacement of the pile body, lateral displacement, etc.

[0071] Vibration data: Vibration response data of the pile and surrounding structure are collected during the pile driving process using accelerometers or vibration sensors to reflect the pile-soil interaction state;

[0072] Stress-strain data: Strain gauges are installed on the steel pipe pile or its auxiliary structure to obtain the stress changes of the pile body;

[0073] Pile hammer impact force data: The impact load generated when the pile hammer comes into contact with the pile body during the pile driving process is recorded in real time by force sensors;

[0074] Pile driving tilt angle data: Use attitude sensors (such as gyroscopes and inclinometers) to monitor the attitude changes of the pile during the pile driving process to ensure construction accuracy;

[0075] Environmental data includes external environmental parameters such as wind speed, wave height, and water depth, which can be acquired in real time through environmental monitoring devices and used to assess the impact of external disturbances on the pile driving process.

[0076] Equipment operating status data, such as hydraulic system pressure and lifting equipment movement status, are collected by sensors in the PLC system or controller to reflect the operating conditions of the pile driving equipment.

[0077] Subsequently, based on the physical characteristics of the pile driving system and the aforementioned multi-source sensor data, a state vector containing multi-dimensional state variables is constructed, represented as:

[0078] ;

[0079] In the formula, For state vectors, Let be the pile's attitude angle vector. The angular velocity vector of the pile body. Let be the acceleration vector of the pile. This refers to the axial displacement of the pile. The axial velocity of the pile body. This is the reaction force of the pile driving. For platform disturbance force, For the total disturbance torque, This represents the environmental disturbance load vector.

[0080] Subsequently, to achieve real-time estimation of state variables, an extended Kalman filter algorithm is used for data fusion and state estimation. The system state is dynamically updated in the recursive process by combining the nonlinear dynamic model and the observation model of the system.

[0081] First, the prior state is predicted using the following state prediction equation:

[0082] ;

[0083] In the formula, Estimate the prior state at the current moment. This is the posterior estimate of the state at the previous time step. To control the input amount.

[0084] After obtaining the prior estimate, the sensor observation data are fused using the following observation update equation:

[0085] ;

[0086] In the formula, To unify the optimal state vector, For prior state estimation, For Kalman gain, For real-time acquisition of multi-source sensor data, To predict the observed values.

[0087] The formula for calculating the Kalman gain matrix is:

[0088] ;

[0089] In the formula, Here is the Kalman gain matrix. Let be the prior state covariance matrix. For the observation matrix, To observe the noise covariance matrix.

[0090] To accurately characterize the stress and motion response of the pile during pile driving, a dynamic model of the pile driving system is established, as described below:

[0091] ;

[0092] In the formula, The equivalent mass of the pile. This is the equivalent damping coefficient. This is the equivalent stiffness coefficient. For pile driving excitation force, For pile-soil reaction force, For platform disturbance force, This refers to environmental disturbance loads.

[0093] In practical applications, the aforementioned dynamic model and state estimation algorithm work together to continuously update the system state through real-time data collected by the sensor network from IMU, DGPS, tilt sensor, strain gauge, anemometer, flow meter, etc., thereby achieving state tracking and identification throughout the pile driving operation.

[0094] This implementation method achieves high-precision estimation of key state variables and dynamic modeling of physical processes in pile driving systems by constructing state vectors, dynamic models, and extended Kalman filter fusion algorithms, forming a core modeling foundation to support subsequent control optimization and decision calculation.

[0095] S3. Based on the unified optimal state vector, the preset coupled system dynamics model is identified and its parameters are updated online to obtain an adaptive dynamics model that can accurately reflect the current real dynamic characteristics of the system.

[0096] Specifically, the actual output at the current moment is first extracted from the unified optimal state vector, and a regression vector containing historical states and inputs is constructed:

[0097] ;

[0098] In the formula, For the regression vector, for The historical state vector at any given time. for The historical state vector at any given time. for Input commands at any time, for Input instructions at any time.

[0099] Calculate the current gain matrix based on the parameter covariance matrix and regression vector from the previous time step:

[0100] ;

[0101] In the formula, Here is the Kalman gain matrix. Let be the parameter covariance matrix of the previous time step. For the regression vector, This is the transpose of the regression vector;

[0102] The key parameter vector is recursively corrected using the following formula:

[0103] ;

[0104] In the formula, This is the updated key parameter vector at the current moment. This is the key parameter vector from the previous time step. Here is the gain matrix. This represents the system's actual output at the current moment. This is the regression vector.

[0105] The updated parameter vector θ(t) will be embedded into the pre-defined coupled system dynamics model:

[0106] ;

[0107] In the formula, It is an adaptive dynamics model. for The system state vector at time t. for Time-based control input, for The model parameter vector at time step, It is a nonlinear function that characterizes the dynamic properties of the system.

[0108] The dynamic model encompasses the pile-soil interaction, the six-degree-of-freedom motion of the hull, and the coupled equations of environmental disturbances. The model is adaptively adjusted through dynamic parameter updates.

[0109] To address nonlinear operating conditions, this implementation can be extended to a recursive least squares method with a forgetting factor, by introducing a forgetting factor. (Typically, a value of 0.95 to 0.99 is used) to suppress data saturation:

[0110] ;

[0111] ;

[0112] In the formula, Here is the Kalman gain matrix. Let be the parameter covariance matrix at the current time. For the regression vector, Let be the parameter covariance matrix of the previous time step. Forgetting factor, This is the transpose of the regression vector.

[0113] The resulting adaptive dynamic model will serve as the core deduction tool for multi-objective optimization control, enabling high-precision prediction and decision-making during the pile driving process.

[0114] This implementation method achieves adaptive online calibration of the dynamic model of the pile driving system by constructing regression vectors, recursive parameter identification algorithms and dynamic covariance update mechanisms, forming a foundation for dynamic modeling of the physical process that accurately reflects the current working conditions, and providing core model support for subsequent control optimization and decision calculation.

[0115] S4. Apply an adaptive dynamics model to perform forward-looking simulations and multi-objective optimization calculations to determine the unique optimal strategic control command that takes into account attitude, energy consumption, and safety.

[0116] Specifically, firstly, the model predicts and controls the MPC framework to achieve forward-looking optimization of the pile driving process.

[0117] The first step of the process is to generate a set of candidate control sequences covering a preset control time domain. Each candidate control sequence consists of a series of control inputs to be applied in future time steps, which aims to explore the possibility of different operation paths.

[0118] Subsequently, in the forward-looking simulation phase, the system takes each candidate control sequence as input and substitutes it into the adaptive dynamic model at the current moment for iterative calculation, thereby deduce the system state trajectory corresponding to the control sequence in the future prediction time domain and predict the possible consequences.

[0119] After obtaining the future state trajectories corresponding to all candidate control sequences, a comprehensive performance evaluation function integrating attitude deviation, energy consumption index, and safety constraints is constructed to quantify the merits of each trajectory. The calculation is achieved through the following formula:

[0120] ;

[0121] In the formula, This is a comprehensive performance evaluation value. To predict discrete time steps in the time domain, To predict the length of the time domain, For the predicted future system state, For the corresponding expected state, For candidate control inputs, The state weight matrix is... To control the input weight matrix, The penalty function for safety constraints.

[0122] Then, the system enters the optimization decision-making stage, the core task of which is to select the unique solution that can make the comprehensive performance evaluation value optimal from many candidate control sequences.

[0123] This optimization process is not unconstrained, but strictly subject to a series of preset physical and safety conditions to ensure that the final generated control commands are engineering feasible. The constraints are as follows:

[0124] Limit the amplitude of control commands to ensure that the commands are within the capabilities of the actuator;

[0125] Set safety boundaries for key state quantities of the pile, such as the maximum allowable inclination angle and displacement, to prevent structural risks.

[0126] The rate of change of control commands is constrained to avoid excessive impact on the equipment and ensure the smoothness of the control process.

[0127] After defining the optimization objective and constraints, the system invokes internally integrated numerical optimization algorithms, such as sequential quadratic programming, to solve the problem. Ultimately, the algorithm calculates a complete control sequence with optimal overall performance within the prediction time domain.

[0128] Following the core idea of ​​rolling optimization, the system does not execute all the instructions in the optimal control sequence, but only extracts and adopts the first control instruction in the sequence as the only optimal strategic control instruction to be output in the current control cycle.

[0129] The instruction is then passed to be parsed and physically executed. Once the instruction is completed, the system enters the next control iteration cycle and restarts the complete decision-making process from deduction, evaluation to optimization based on the updated actual state information.

[0130] This implementation constructs a model predictive control framework based on an adaptive model, uses a comprehensive performance evaluation function to prospectively assess the future multi-step control effect, and performs rolling optimization under multiple engineering constraints. This achieves an advanced control strategy that can dynamically adapt to system changes and balance multiple control objectives, providing core algorithmic support for ensuring high precision, high efficiency, and high safety in pile driving operations.

[0131] S5. Perform multi-scale collaborative analysis and physical execution of the optimal strategic control command, and initiate the next round of control iteration through a closed-loop feedback mechanism;

[0132] Specifically, firstly, the system needs to perform collaborative analysis of the optimal strategic control command. This command, as a macro-level decision, needs to be precisely decomposed into multiple sub-commands targeting different physical execution agencies.

[0133] This decomposition process is implemented through a preset control allocation matrix, and the calculation formula is as follows:

[0134] ;

[0135] In the formula, A vector containing all sub-instructions. This is the optimal strategic control instruction. To control the allocation matrix.

[0136] The matrix is ​​pre-calibrated based on the physical characteristics and layout of the actuators. After obtaining each sub-instruction, the system needs to calculate the underlying drive control signals used to directly drive each physical actuator.

[0137] To ensure fast and error-free tracking of sub-instructions, this invention preferably employs a proportional-integral-derivative (PID) control algorithm in the control loop of each actuator, calculated using the following formula:

[0138] ;

[0139] In the formula, These are the underlying drive control signals. This is the deviation between the target value of the sub-instruction and the feedback value of the physical actuator. For proportional gain, For integral gain, This is the differential gain.

[0140] After the calculation is completed, the underlying drive control signals are sent to the corresponding actuators via industrial fieldbus (such as CAN or Profinet).

[0141] The hammering control signal is sent to the PLC controller of the hydraulic hammer to achieve precise control of the hammering energy and rhythm. The platform control signal is sent to the ship's DP system to suppress environmental disturbances in real time.

[0142] This physical execution will cause a change in the physical state of the entire pile driving operation system. This change in state will be captured in real time by the multi-source sensor network deployed by the system, forming a new round of dynamic information.

[0143] The dynamic information is then used for a new round of state fusion processing to update the state perception of the entire system. The updated state will serve as the starting point to drive the system to make model updates and optimization decisions again, thus forming a complete closed-loop feedback.

[0144] This implementation method constructs an implementation path from macro-strategic instructions to micro-physical execution, and ensures the implementation and iterative cycle of the entire control method through closed-loop feedback, which is a key link in realizing system intelligence.

[0145] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for controlling the sinking of a steel pipe pile for a container wharf in a deep water port area, characterized by, The method comprises the following steps: S1, information deployment of a multi-source sensor network is performed on an initial pile sinking platform to construct an informationized pile sinking system capable of digital sensing; S2, based on the informationized pile sinking system, real-time collection and fusion processing are performed on multi-source sensing data in the pile sinking process to generate a unified optimal state vector representing dynamic characteristics of the system at the current time; S3, based on the unified optimal state vector, online identification and parameter updating are performed on a preset coupled system dynamics model to obtain an adaptive dynamics model accurately reflecting real dynamic characteristics of the system; S4, the adaptive dynamics model is applied to perform forward-looking deduction and multi-objective optimization calculation to determine a unique optimal strategic control instruction considering attitude, energy consumption and safety; S5, the optimal strategic control instruction is subjected to multi-scale collaborative analysis and physical execution, and the next round of control iteration is started through a closed-loop feedback mechanism. The information deployment of the multi-source sensor network on the initial pile sinking platform in the step S1 comprises: an inertial measurement unit and a differential global positioning system are installed on the top of the steel pipe pile, and a plurality of inclination sensors and a plurality of strain gauges are arranged along the length direction of the pile body; a second group of inertial measurement units and a second group of differential global positioning systems for monitoring the attitude of the construction ship are installed, and an environment monitoring device for obtaining wind, wave and flow environmental load data is integrated; The online identification and parameter updating of the preset coupled system dynamics model in the step S3 comprises: the actual output at the current time is extracted from the unified optimal state vector, and a regression vector containing the historical state and input of the system is constructed; a gain matrix at the current time is calculated based on the parameter covariance matrix at the last time and the regression vector; the key parameter vector at the last time is corrected by using the gain matrix and the actual output at the current time of the system to obtain an updated key parameter vector at the current time, and the updated key parameter vector is applied to the preset coupled system dynamics model to obtain the adaptive dynamics model; The forward-looking deduction and multi-objective optimization calculation of the adaptive dynamics model in the step S4 comprises: based on the adaptive dynamics model, a group of candidate control sequences covering the preset control time domain are generated, and the future system state trajectory corresponding to each candidate control sequence is deduced; a comprehensive performance evaluation function integrating attitude deviation, energy consumption index and safety constraint is constructed, and the function is used to quantitatively evaluate each future system state trajectory to obtain its performance evaluation value; the candidate control sequence with the optimal performance evaluation value is selected, and the initial control instruction thereof is taken as the unique optimal strategic control instruction.

2. The method according to claim 1, wherein, The fusion processing of the collected multi-source sensing data in the step S2 comprises: based on the system state estimation and state covariance obtained in the last control cycle, a group of sigma sampling points carrying deterministic weights are generated; the sigma sampling points are substituted into the preset nonlinear system state equation and observation equation respectively for propagation to calculate the prior state estimation and predicted observation value at the current time; the Kalman gain is calculated by combining the real-time collected multi-source sensing data and the predicted observation value; The Kalman gain is used to correct the prior state estimation to generate a unified optimal state vector.

3. The method according to claim 2, wherein, The calculation of the Kalman gain correcting the prior state estimation is implemented by the following formula: ； wherein, is a unified optimal state vector, is a prior state estimate, is a Kalman gain, is real-time collected multi-source sensing data, is a predicted observation.

4. The method according to claim 1, wherein, The calculation of the key parameter vector of the last time being corrected is implemented by the following formula: ； wherein is the updated key parameter vector at the current time instant, is the key parameter vector at the previous time instant, is the gain matrix, is the actual output of the system at the current time instant, is the regression vector.

5. The method for controlling the sinking of steel pipe piles for a deepwater port container terminal according to claim 1, wherein The calculation of the comprehensive performance evaluation function is implemented by the following formula: ； wherein is a comprehensive performance evaluation value, is a discrete time step within the prediction horizon, is the prediction horizon length, is the predicted future system state, is the corresponding desired state, is a candidate control input, is a state weight matrix, is a control input weight matrix, is a penalty function for safety constraints.

6. The method for controlling the sinking of steel pipe piles for a deepwater port container terminal according to claim 1, wherein The multi-scale collaborative analysis and physical execution step of the optimal strategic control instruction in step S5 is: The unique optimal strategic control instruction is decomposed into multiple sub-instructions acting on different time scales and space scales; According to the multiple sub-instructions, the bottom layer driving control signal for driving the physical execution mechanism is calculated and issued to collaboratively complete the pile sinking operation.

7. The method according to claim 6, wherein, The calculation of the bottom layer driving control signal is implemented by the following control algorithm: ； wherein is a bottom layer drive control signal, is a deviation between a target value of a sub-instruction and a feedback value of a physical actuator, is a proportional gain, is an integral gain, is a differential gain.

Images