Crane outrigger predictive leveling method and system based on digital twinning

Abstract

The application discloses a crane outrigger predictive leveling method and system based on digital twinning, comprising: constructing a digital twin model; capturing operation instructions in real time and analyzing, extracting future action parameters representing future operation intentions; taking the current whole machine state and the future action parameters as inputs, driving the reduced order model in the digital twin model to perform fast forward simulation, deducing the dynamic response state sequence of the whole machine in the future prediction time domain, and calculating the future stability margin according to the dynamic response state sequence; obtaining the ground subsidence trend prediction result of each outrigger; fusing the future stability margin and the ground subsidence trend prediction result, making a hierarchical decision according to the comparison between the fusion result and the preset multi-level safety threshold; generating and executing predictive outrigger fine tuning instructions; closed-loop feedback and model iteration. The application can realize the fundamental change of intelligent leveling control mode, establish deep coupling between operation instructions and outrigger control, and improve the dynamic adaptation capability to complex environments.

Description

Technical Field

[0001] This invention relates to a method and system for leveling crane outriggers, and more particularly to a predictive leveling method and system for crane outriggers based on digital twins. Background Technology

[0002] As an indispensable large-scale mechanical equipment in modern engineering construction, the safe operation of cranes is the fundamental guarantee for the smooth progress of projects. The outrigger system, as the supporting foundation of the crane, directly determines the stability and safety of the entire machine's operation through its leveling accuracy, response speed, and dynamic adaptability.

[0003] Currently, the outrigger leveling technology of mainstream cranes has evolved from manual to automatic, and can be mainly divided into the following technical stages:

[0004] Phase 1: Manual Leveling

[0005] Operators manually operate hydraulic valves to control the extension and retraction of each outrigger by visually inspecting a spirit level or a simple bubble level until the level is visually apparent. This method is inefficient, inaccurate, heavily reliant on operator experience, and cannot be dynamically adjusted during operation.

[0006] Phase Two: Sensor-Based Automatic Leveling

[0007] A typical example is the "Automatic Leveling Control System for Crane Outriggers" disclosed in patent document CN201120571821.2. This system uses dual-axis tilt sensors installed on the vehicle body to detect the platform's levelness in real time. Based on the tilt deviation, the controller automatically controls the hydraulic valve group according to preset control logic (such as the highest point stationary method or the unidirectional successive approximation method), driving the corresponding outriggers to move until the platform reaches a set levelness threshold. This type of method automates the leveling process, significantly improving leveling efficiency and initial accuracy.

[0008] Phase 3: Intelligent leveling with pressure monitoring

[0009] Building upon the second stage, outrigger pressure sensors are added, such as the "crane outrigger pressure and verticality detection device" described in patent document CN200820176413.5. This system not only levels the crane but also monitors the stress on each outrigger, preventing "loose legs" (outriggers suspended in mid-air), and makes certain compensation adjustments based on pressure changes during lifting.

[0010] In the prior art, CN115783996A discloses an automatic leveling system for crane outriggers, which uses a fuzzy PID algorithm to control the solenoid valves, improving leveling speed and accuracy. However, this still falls under the category of reactive feedback control and does not incorporate simulation prediction of future operating conditions, thus failing to address the lag problem.

[0011] CN112010195A describes a crane outrigger coordinated control system based on tilt and pressure sensors, which achieves coordinated control of the outriggers. While focusing on the coordinated movement of multiple outriggers, the control relies on real-time feedback signals and lacks the ability to predict future risks.

[0012] CN117049379B describes a tower crane operation status monitoring and evaluation system based on digital twins. While it utilizes digital twins, its primary application is in monitoring, evaluating, and providing collision avoidance warnings for operational status. It does not involve active or predictive control of outrigger leveling, and the digital twin model and actuators do not form a control closed loop.

[0013] Despite significant advancements in existing technology, the following fundamental shortcomings remain:

[0014] (1) The lag in control – the essence of “reactive” regulation

[0015] All existing technologies are essentially "feedback control," meaning that deviations occur first, and corrections follow. Whether it's initial leveling or compensation during operation, the actuators are driven based on tilt or pressure deviations detected "at the current moment." This "post-event adjustment" mode has an inherent time lag, making it impossible to anticipate and proactively avoid the instantaneous instability risks caused by rapid load transfers and sudden changes in the center of gravity during dynamic operations such as lifting, slewing, and luffing of cranes.

[0016] (2) One-sidedness in information utilization – disconnected from future working conditions

[0017] Existing leveling systems make control decisions based solely on current state data (current tilt angle, current pressure), completely disconnected from the "operational instructions to be executed." For example, when the operator pushes the control lever to prepare for a large-scale rotation, the leveling system is completely unaware of this "future operating condition" and continues to maintain the current levelness step by step. The system only begins to react after the rotation causes the vehicle to tilt out of tolerance, by which time a danger may have already occurred.

[0018] (3) Insufficient environmental adaptability – unable to predict ground subsidence

[0019] Existing systems are ineffective in addressing changes in the supporting ground. When a crane operates on soft soil or backfill, the foundation beneath the outriggers may experience continuous settlement during lifting. This settlement is a gradual process, and existing systems can only correct it after the settlement has caused the crane body to tilt. They cannot predict the settlement trend and take compensatory measures in advance before or at the beginning of lifting.

[0020] (4) Limitations of human-computer interaction – information black box

[0021] Operators are typically faced with simple digital instruments or indicator lights, which only provide fragmented information such as "whether the current level is correct" and "whether the current pressure exceeds the limit." They cannot intuitively and comprehensively understand the stability margin of the entire machine, future risk trends, and the force reserves of each outrigger, resulting in insufficient basis for decision-making. Summary of the Invention

[0022] Purpose of the invention: The purpose of this invention is to propose a predictive leveling method and system for crane outriggers based on digital twins, to achieve a fundamental shift in intelligent leveling control mode, establish deep coupling between operating commands and outrigger control, and improve the dynamic adaptability to complex environments.

[0023] Technical solution: This invention includes:

[0024] Construct a digital twin model that is synchronized with the physical crane in real time;

[0025] Real-time capture and parsing of operation commands, extracting future action parameters that represent future task intentions;

[0026] Using the current state of the whole machine and future action parameters as input, the reduced-order model in the digital twin model is driven to perform rapid forward simulation, deduce the dynamic response state sequence of the whole machine in the future prediction time domain, and calculate the future stability margin based on the dynamic response state sequence.

[0027] Obtain the predicted ground settlement trend results for each leg;

[0028] The system integrates future stability margin with land subsidence trend prediction results, and makes graded decisions on maintaining silence, triggering early warning, or initiating active intervention based on the comparison between the integrated results and preset multi-level safety thresholds.

[0029] When active intervention is initiated, a model predictive control algorithm is used to solve for the optimal control sequence that maximizes future stability margin under the premise of satisfying preset constraints, and the control sequence is executed to dynamically fine-tune the outriggers.

[0030] The fine-tuned actual state is fed back to the digital twin model for state updates, and the reduced-order model and the ground subsidence model are iteratively optimized online using prediction-decision-execution process data.

[0031] The order reduction model is a deep learning-based model that pre-generates a sample dataset covering various working conditions using simulation software and trains it offline. It is used to express the nonlinear mapping relationship between the current state and future operations as a sequence of future states.

[0032] The land subsidence trend prediction specifically includes:

[0033] Ground characteristic identification: During the initial grounding and preloading process of the outriggers, the system records the outrigger displacement change curve as the outrigger pressure increases from zero to the preset preload value;

[0034] Constructing a settlement prediction model: For support points identified as soft ground, establish a settlement prediction equation based on a creep model. Substitute the predicted future time points into the model to obtain the dynamic settlement prediction of each leg under future working conditions.

[0035] The multi-level security thresholds include:

[0036] First safety threshold: Indicates that the future safety margin is low, and a prompt will be provided through the human-computer interaction terminal;

[0037] The second safety threshold indicates that the future safety margin is approaching the danger zone, and the system must actively intervene. When the future stability margin is lower than the second safety threshold, the system will bypass the warning and directly start the predictive leveling strategy generation process.

[0038] The future stability margin includes the minimum reaction force margin of the outriggers, the overturning stability margin, and the comprehensive stability margin index. The comprehensive stability margin index is calculated by weighting the minimum reaction force margin of the outriggers and the overturning stability margin.

[0039] The preset constraints include:

[0040] Kinematic constraints: The outrigger's fine-tuning movements must not interfere with the main operation being performed by the upper garment;

[0041] Actuator constraints: The extension and retraction speed and range of the outriggers are within physical limits;

[0042] Stability constraint: During fine-tuning, the overall stability margin index at any time must not be lower than the second safety threshold.

[0043] The present invention also provides a predictive leveling system for crane outriggers based on digital twins, comprising:

[0044] The data acquisition and edge computing module, deployed at the physical crane end, includes sensors, signal conditioning circuits, and an edge computing unit. This unit is responsible for the real-time acquisition and preprocessing of sensor data and the execution of instructions from the main controller.

[0045] The digital twin construction and simulation module includes a model management unit, a real-time synchronization unit, and a simulation and deduction unit;

[0046] The prediction and decision control module includes a stability margin calculation unit, a ground settlement prediction unit, and an optimization decision unit;

[0047] The execution control module interfaces with the existing outrigger hydraulic control system, receives fine-tuning instructions from the decision control module, and converts them into control current for proportional valves or servo valves to realize outrigger movements.

[0048] The digital twin construction and simulation module is deployed on an in-vehicle industrial control computer or a cloud server.

[0049] The model management unit is used to store and manage the geometric model, physical model, and dynamic model of the crane; the real-time synchronization unit is used to receive real-time data streams and drive the model to keep synchronized with the physical entity; the simulation derivation unit has a built-in trained reduced-order model and is responsible for performing fast forward simulation.

[0050] The stability margin calculation unit calculates the SMI index based on the simulation results; the ground settlement prediction unit runs the ground settlement model to predict the settlement trend; and the optimization decision unit runs the model predictive control algorithm to solve for the optimal outrigger fine-tuning command.

[0051] Beneficial effects: This invention has the following advantages:

[0052] (1) Improve the safety of construction machinery operation: This invention fundamentally solves the problem of lag in traditional leveling systems. It can predict and actively eliminate the risk of instability caused by center of gravity shift, ground settlement, sudden wind changes, etc., which can significantly reduce the incidence of crane overturning accidents and ensure the safety of personnel and equipment. According to industry statistics, about 30% of crane accidents are related to outrigger instability. The promotion and application of this invention is expected to reduce the risk of such accidents by more than 50%.

[0053] (2) Promoting the intelligent upgrading of construction machinery: This invention deeply integrates cutting-edge technologies such as digital twins, deep learning, and model predictive control, providing a model for the crane and even the entire construction machinery industry to leap from "automation" to "intelligentization". Its core technology can be extended to the stability control of various engineering equipment such as concrete pump trucks, aerial work platforms, and crawler cranes, forming a series of intelligent control products.

[0054] (3) Improve operational efficiency and lower the operational threshold: Through predictive leveling and human-machine collaborative decision-making, manual intervention is reduced, freeing operators from tedious stability monitoring and allowing them to focus on the hoisting operation itself. At the same time, the immersive visual interface lowers the operational experience threshold, making it easier for novices to get started and helping to alleviate the shortage of skilled operators.

[0055] (4) Huge Market Value: With the continuous advancement of global infrastructure construction, the number of cranes in operation is steadily increasing, and the demand for intelligent transformation of existing equipment and intelligent configuration of new equipment is strong. This invention can serve as a core selling point for high-end cranes, enhancing product added value and strengthening the international competitiveness of enterprises. It is estimated that the market size of related intelligent control systems will reach tens of billions of yuan in the next five years.

[0056] (5) Strong technical scalability: The digital twin and predictive control platform constructed by this invention can be seamlessly connected with the construction management system and remote operation and maintenance platform to realize value-added services such as equipment health management and operation optimization scheduling, opening up new space for the aftermarket service of construction machinery. Attached Figure Description

[0057] Figure 1 This is a flowchart of the present invention;

[0058] Figure 2 This is a diagram showing the module composition and data flow of the predictive leveling system of the present invention.

[0059] Figure 3 This is a schematic diagram of the core logic for prediction and optimization in this invention;

[0060] Figure 4 This is a schematic diagram of the hierarchical decision-making logic and immersive visualization interface of the present invention. Detailed Implementation

[0061] The invention will now be further described with reference to the accompanying drawings.

[0062] Example 1

[0063] like Figure 1 As shown in this embodiment, the predictive leveling method for crane outriggers based on digital twins upgrades the traditional "reactive leveling" to "predictive stability maintenance," significantly improving the operational safety of cranes under complex dynamic conditions. Specifically, it includes the following steps:

[0064] S1. Real-time acquisition of crane operation and environmental data through a multi-source sensor network, construction of a digital twin model that is precisely synchronized with the physical entity, and realization of virtual-real mapping of geometric posture and mechanical behavior;

[0065] Specifically, based on the precise geometric dimensions, mass distribution, moment of inertia, and hydraulic system response characteristics of the physical crane, as well as its 3D CAD model, a virtual model is constructed that is highly consistent with the physical entity in terms of structure and behavior. Simultaneously, a multi-source sensor network is deployed at key parts of the physical crane to collect operational data in real time and synchronously drive the virtual model, achieving virtual-real synchronization.

[0066] Multi-source sensor networks specifically include:

[0067] The displacement sensor of the outrigger vertical cylinder monitors the outrigger extension and retraction in real time.

[0068] The pressure sensor in the rodless chamber of the outrigger vertical cylinder monitors the outrigger support reaction force in real time.

[0069] The dual-axis tilt sensor of the vehicle body slewing platform monitors the longitudinal and lateral levelness of the platform in real time;

[0070] Angle and length sensors at the base and top of the boom monitor the boom's attitude in real time.

[0071] Weight sensors on hooks or hoisting winches monitor the lifting load in real time;

[0072] An anemometer at the highest point of the vehicle body monitors environmental wind loads in real time.

[0073] Strain sensors for critical structural components (such as booms and frames) are used to monitor the stress state of the structure in real time.

[0074] The virtual-real synchronization specifically involves using the aforementioned sensor data as input to drive a geometric model built on a physics engine (such as Unity3D or Unreal Engine) to achieve precise attitude mapping. Based on a multibody dynamics model and combined with real-time load data, the real-time center of gravity position of the entire machine and the force state of each component are calculated to achieve precise mapping of mechanical behavior. The data synchronization frequency is no less than 50Hz to ensure real-time consistency between the digital twin model and the physical entity.

[0075] S2. Receive and parse operation commands in real time, extract future action parameters (crane's conventional action types, boom extension and retraction, winch lifting and lowering, luffing lifting and lowering, left and right slewing, control speed, and load), drive the digital twin model to use the whole machine design model, parameter presets, and intelligent algorithm simulation to deduce the dynamic response of the whole machine in the future prediction time domain, and calculate the comprehensive stability margin index.

[0076] Specifically, this involves real-time capture of operator commands issued via joysticks, remote controls, or touchscreens, as well as preset automated task commands. The command parsing module converts the raw command signals into quantified future action parameters that the simulation model can understand. These future action parameters include:

[0077] Action types: hoisting / lowering, boom extension / retraction, left / right rotation, luffing / lowering, and their combined actions.

[0078] Range of motion: target range, target rotation angle, target lifting height, etc.

[0079] Action speed / acceleration: The desired speed and acceleration curves corresponding to the lever opening.

[0080] Expected load: Current lifting load, and the change in load amplitude during luffing.

[0081] S3. Future working condition simulation and deduction based on digital twin;

[0082] like Figure 3As shown, the future motion parameters analyzed by S2 are used as excitation inputs to the digital twin model. Utilizing a high-performance simulation engine deployed on an edge computing unit or in the cloud, the digital twin model is driven to perform rapid forward simulation in virtual space, simulating the dynamic response of the crane over a period of time (e.g., 10-30 seconds in the future, with the prediction time domain dynamically adjustable via a human-machine interface terminal) starting from the current moment after the execution of the instruction.

[0083] The core algorithm of simulation and deduction—the reduced-order model (ROM):

[0084] To meet real-time requirements (each simulation must be completed within milliseconds), complex full-order finite element or multibody dynamics models cannot be directly used. This invention employs a reduced-order model based on deep learning.

[0085] (1) Model building: A large number of offline simulations were performed in advance using high-precision simulation software (such as Adams and Ansys) to generate a massive "condition-response" dataset covering various working conditions (different lifting weights, amplitudes, boom lengths, and ground stiffness). A deep neural network (DNN) was trained using this dataset to learn the nonlinear mapping relationship from "current state + future operation" to "future state sequence".

[0086] (2) Model input: [Current whole machine state vector S_t] + [Future action parameter vector A_t].

[0087] (3) Model output: [Predicted future state vector sequence S_(t+1), S_(t+2), ..., S_(t+n)], where the state vector includes at least: the position of the center of gravity of the whole machine, the reaction force of each outrigger, the platform tilt angle, and the overturning moment.

[0088] (4) Calculate the future stability margin:

[0089] Minimum reaction margin for outriggers: M_f = min(F_i) / F_nom, where F_i is the predicted reaction force for each outrigger and F_nom is the rated reaction force for the outriggers. This value reflects the risk of the outriggers lifting off the ground; the lower the value, the greater the risk.

[0090] Overturning stability margin: M_o = 1 - M_t / M_rated, where M_t is the predicted total overturning moment (including the moment generated by self-weight, hoisting load, wind load, inertial force, etc.), and M_rated is the rated overturning moment. This value reflects the overall risk of overturning; the lower the value, the greater the risk.

[0091] The Comprehensive Stability Margin Index (SMI) can be defined as a weighted index that takes into account both of the above margins, for example, SMI = α * M_f + β * M_o, and is used for subsequent decision-making.

[0092] S4: Ground Subsidence Trend Prediction

[0093] To address the uncertainties in the ground support provided by cranes, a data-driven ground settlement prediction model is constructed.

[0094] (1) Ground feature identification:

[0095] During the initial grounding and preloading process of the outriggers, the system records the outrigger displacement curve as the outrigger pressure increases from zero to the preset preload value. Ground with large displacement changes and slow response is identified as soft ground; ground with small displacement changes and fast response is identified as hard ground.

[0096] During pressure holding and operation, the displacement changes of each outrigger under constant control pressure are continuously monitored. If a continuous, monotonically increasing trend in displacement is detected (i.e., the outrigger slowly sinks), it is determined that there is settlement at that support point, and its settlement rate is identified.

[0097] (2) Constructing a settlement prediction model:

[0098] For support points identified as soft ground, settlement prediction equations based on creep models are established, such as Δh(t) = k * ln(1 + t / τ) or other empirical formulas, where k and τ are parameters related to ground characteristics, and are identified and updated online using real-time monitoring data.

[0099] By substituting the predicted future time points into the model, the predicted dynamic settlement of each leg under future working conditions, Δh_pred(t), can be obtained.

[0100] S5. Integrate future stability margins with settlement prediction results, compare them with preset multi-level safety thresholds, and make graded decisions such as maintaining the status quo, triggering early warnings, or initiating active intervention. Figure 4 As shown; when the decision is to actively intervene, the model predictive control (MPC) method is adopted. Under the constraint of ensuring the continuity of the main operation of the superstructure, the optimal outrigger fine-tuning control algorithm that maximizes the future stability margin is solved and executed.

[0101] (1) Define multiple security thresholds:

[0102] First safety threshold (early warning threshold): For example, if the comprehensive stability margin index SMI < 0.3, it indicates that the future safety margin is low and attention should be paid. This will be indicated through the human-computer interaction terminal.

[0103] The second safety threshold (intervention threshold): For example, if the comprehensive stability margin index SMI < 0.15, it means that the future safety margin is approaching the danger zone, and the system must actively intervene.

[0104] (2) Decision-making logic:

[0105] Scenario 1 (Safety): If the overall stability margin index SMI is always higher than the first safety threshold throughout the entire prediction time domain, and the predicted settlement Δh_pred(t) of each outrigger is less than the allowable settlement Δh_allow, then the future working condition is determined to be safe, the system does not trigger any action, and remains in a silent monitoring state.

[0106] Scenario 2 (Warning): If the Comprehensive Stability Margin Index (SMI) falls below the first safety threshold but remains above the second safety threshold at some point in the future, or if the predicted settlement of a certain outrigger exceeds the allowable value, the system will trigger a Level 1 warning. The risk area will be highlighted on the human-machine interface, and the operator will be prompted with a voice message, "Attention, stability of the working condition ahead is decreasing," while an optimization suggestion will be generated (such as "It is recommended to reduce the slewing speed" or "It is recommended to make minor adjustments to the outriggers first").

[0107] Scenario 3 (Proactive Intervention): If the Comprehensive Stability Margin Index (SMI) falls below the second safety threshold at some point in the future, the system determines that there is a high risk and immediately triggers Level 2 proactive intervention. In this case, the system will bypass the early warning and directly initiate the predictive balancing strategy generation process.

[0108] S6. Generate and execute predictive outrigger fine-tuning instructions.

[0109] Once proactive intervention is triggered or the operator confirms and adopts the warning suggestion, the system starts the optimization algorithm to find the optimal outrigger fine-tuning solution.

[0110] (1) Establish the optimization objective function:

[0111] J = max_{u(t)} [ min_{t∈[T, T+Δt]} SMI(t) ], which means maximizing the overall stability margin at the worst-case moment over a future period. In other words, it seeks a set of support leg fine-tuning actions to maximize the improvement of the "weakest link" in security throughout the entire prediction time domain.

[0112] (2) Constraints:

[0113] a. Kinematic constraints: The outrigger fine-tuning motion must not interfere with the main operation of the upper structure (such as rotation or lifting). Decoupling control is usually adopted, which decomposes the fine-tuning motion into components that are orthogonal to the main motion or have the least impact, or utilizes the interval of the main motion to complete it quickly.

[0114] b. Actuator constraints: The extension and retraction speed and range of the outriggers are within the physical limits.

[0115] c. Stability Constraint: During fine-tuning, the overall stability margin index (SMI) must not fall below the second safety threshold at any time.

[0116] (3) Solving and execution:

[0117] A Model Predictive Control (MPC) framework is used to solve the above optimization problem in a rolling manner. In each control cycle, based on the latest state and predictions, the optimal control sequence (the expected extension and contraction speed of each leg) for a shorter future time domain (e.g., 2 seconds) is calculated.

[0118] The first value of the control sequence is used as the control command for the current cycle and sent to the electro-hydraulic proportional valve of the outrigger, enabling continuous and smooth fine-tuning of the outrigger. The entire process may be "transparent" to the operator, who can hardly feel the movement of the outrigger, but the overall stability of the vehicle is improved.

[0119] S7, Closed-Loop Feedback and Model Iteration

[0120] The actual effects of the outrigger fine-tuning (measured by sensors, including changes in tilt angle and pressure) are fed back to the digital twin model in real time to update its status and evaluate the actual effect of this predictive leveling.

[0121] The entire process of prediction, decision-making, and execution (including initial predictions, actual results, and errors) is stored in a database for incremental learning and online updates of the reduced-order model and the ground subsidence model. This allows the model's prediction accuracy to continuously improve with increasing usage, thus enabling the system to self-evolve.

[0122] Step S8: Immersive Visual Interaction

[0123] Build an immersive digital twin visualization interface on human-computer interaction terminals (such as touchscreens in the driver's cab, displays on remote controls, etc.). This interface should include:

[0124] Current status visualization: Display the crane's posture in real time using a 3D model.

[0125] Visualization of future states: The predicted future posture and motion trajectory are displayed by overlaying semi-transparent models or trajectory lines.

[0126] Safety situation visualization: The magnitude of force is displayed at each outrigger location in the form of a heat map (green for safety, yellow for warning).

[0127] (Red danger) Draw a stability margin cloud map or safety bubble around the entire machine to visually show the safe area and the direction of risk.

[0128] Trend curve visualization: Real-time plotting of the predicted change curve of the stability margin (SMI) over a future period, allowing for operational...

[0129] Employees can anticipate the evolving trends of risks.

[0130] Visualized suggestions and tips: Display system-provided optimization suggestions or warning messages in the form of text, icons, or voice.

[0131] This invention enables a fundamental shift in intelligent leveling control mode: transforming traditional "reactive control based on current deviation" into "feedforward-feedback composite control based on future prediction," fundamentally solving the problem of control lag; establishing deep coupling between operating commands and outrigger control: enabling the outrigger control system to "understand" the operator's intentions and pre-assess the impact of upcoming operations on the overall stability of the machine, thus providing parameter support for pre-control; enhancing dynamic adaptability to complex environments: achieving proactive compensation by predicting environmental factors such as ground subsidence, reducing the impact of environmental interference; and constructing intelligent human-machine interaction: through immersive visualization, enabling operators to intuitively grasp the safety situation and achieve collaborative decision-making between humans and the intelligent system.

[0132] Example 2

[0133] like Figure 2 As shown, the predictive leveling system for crane outriggers based on digital twins in this embodiment includes the following modules:

[0134] Data Acquisition and Edge Computing Module: Deployed at the physical crane end, this module includes all the aforementioned sensors, signal conditioning circuitry, and an edge computing unit. This unit is responsible for real-time acquisition and preprocessing (filtering, synchronization) of sensor data, as well as executing instructions from the main controller.

[0135] Digital twin construction and simulation module: Can be deployed on a high-performance industrial control computer in a vehicle or a cloud server. This module includes:

[0136] Model management unit: Stores and manages the geometric model, physical model, and dynamic model of the crane.

[0137] Real-time synchronization unit: Receives real-time data streams and drives the model to keep synchronized with the physical entity.

[0138] Simulation and derivation unit: Built-in pre-trained reduced-order model (deep neural network) is responsible for performing fast forward simulation.

[0139] Predictive and Decision Control Module: The system's "intelligent brain," deployed in the vehicle's main controller. This module includes:

[0140] Stability margin calculation unit: Calculates indices such as SMI based on simulation results.

[0141] Ground subsidence prediction unit: Runs the ground subsidence model to predict subsidence trends.

[0142] Optimization Decision Unit: Run the Model Predictive Control (MPC) algorithm to solve for the optimal outrigger fine-tuning command.

[0143] Execution control module: Interfaces with the existing outrigger hydraulic control system, receives fine-tuning instructions from the decision control module, and converts them into control current for proportional valves or servo valves to achieve precise outrigger movements.

[0144] Human-computer interaction module: including graphic display, voice input / output device, AR / VR interface, etc., used to realize immersive visual interaction and receive operator instructions.

[0145] This invention upgrades the traditional "reactive leveling" to "predictive stability maintenance." Based on digital twin simulation to predict future working conditions, it proactively adjusts the outriggers before risks occur, fundamentally solving the problem of lag and elevating safety protection from "post-event remediation" to "pre-event prevention." It establishes deep coupling between operating commands and outrigger control. By analyzing operating commands in real time, the outrigger system "understands" the operator's intentions, pre-assesses the impact of actions such as slewing and lifting on stability, and dynamically adjusts accordingly. It breaks the traditional situation of independent systems, achieving integrated intelligent control of the entire machine. It enhances the dynamic adaptability to complex environments. Based on pressure-displacement data, it identifies ground stiffness, predicts settlement in soft soil foundations, and proactively compensates for it. It eliminates environmental interference, significantly improving the safety of crane operations in harsh conditions such as backfill soil and soft soil. It constructs an immersive interactive interface for human-machine collaboration. It intuitively displays the safety situation in the form of 3D heat maps, safety cloud maps, and trend curves, allowing operators to intuitively grasp current and future risks. It realizes the graphical representation of abstract data, lowers the operational threshold, and forms collaborative decision-making between humans and intelligent systems. This forms a self-evolving intelligent closed-loop system. The prediction, decision-making, and execution data from each task are used for incremental model learning, continuously improving prediction accuracy with each use and fine-tuning the strategy for continuous optimization. This achieves a self-evolving capability that "gets smarter with use," increasing the technological value over time.

Claims

1. A predictive leveling method for crane outriggers based on digital twins, characterized in that, Includes the following steps: Construct a digital twin model that is synchronized with the physical crane in real time; Real-time capture and parsing of operation commands, extracting future action parameters that represent future task intentions; Using the current state of the whole machine and future action parameters as input, the reduced-order model in the digital twin model is driven to perform rapid forward simulation, deduce the dynamic response state sequence of the whole machine in the future prediction time domain, and calculate the future stability margin based on the dynamic response state sequence. Obtain the predicted ground settlement trend results for each leg; The system integrates future stability margin with land subsidence trend prediction results, and makes graded decisions on maintaining silence, triggering early warning, or initiating active intervention based on the comparison between the integrated results and preset multi-level safety thresholds. When active intervention is initiated, a model predictive control algorithm is used to solve for the optimal control sequence that maximizes future stability margin under the premise of satisfying preset constraints, and the control sequence is executed to dynamically fine-tune the outriggers. The fine-tuned actual state is fed back to the digital twin model for state updates, and the reduced-order model and the ground subsidence model are iteratively optimized online using prediction-decision-execution process data.

2. The predictive leveling method for crane outriggers based on digital twins according to claim 1, characterized in that, The order reduction model is a deep learning-based model that pre-generates a sample dataset covering various working conditions using simulation software and trains it offline. It is used to express the nonlinear mapping relationship between the current state and future operations as a sequence of future states.

3. The predictive leveling method for crane outriggers based on digital twins according to claim 1, characterized in that, The land subsidence trend prediction specifically includes: Ground characteristic identification: During the initial grounding and preloading process of the outriggers, the system records the outrigger displacement change curve as the outrigger pressure increases from zero to the preset preload value; Constructing a settlement prediction model: For support points identified as soft ground, establish a settlement prediction equation based on a creep model. Substitute the predicted future time points into the model to obtain the dynamic settlement prediction of each leg under future working conditions.

4. The predictive leveling method for crane outriggers based on digital twins according to claim 1, characterized in that, The multi-level security thresholds include: First safety threshold: Indicates that the future safety margin is low, and a prompt will be provided through the human-computer interaction terminal; The second safety threshold indicates that the future safety margin is approaching the danger zone, and the system must actively intervene. When the future stability margin is lower than the second safety threshold, the system will bypass the warning and directly start the predictive leveling strategy generation process.

5. The predictive leveling method for crane outriggers based on digital twins according to claim 1, characterized in that, The future stability margin includes the minimum reaction force margin of the outriggers, the overturning stability margin, and the comprehensive stability margin index. The comprehensive stability margin index is calculated by weighting the minimum reaction force margin of the outriggers and the overturning stability margin.

6. The predictive leveling method for crane outriggers based on digital twins according to claim 1, characterized in that, The preset constraints include: Kinematic constraints: The outrigger's fine-tuning movements must not interfere with the main operation being performed by the upper garment; Actuator constraints: The extension and retraction speed and range of the outriggers are within physical limits; Stability constraint: During fine-tuning, the overall stability margin index at any time must not be lower than the second safety threshold.

7. A predictive leveling system for crane outriggers based on digital twins, characterized in that, The method for performing the predictive leveling of crane outriggers based on digital twins as described in any one of claims 1 to 6 includes: The data acquisition and edge computing module, deployed at the physical crane end, includes sensors, signal conditioning circuits, and an edge computing unit. This unit is responsible for the real-time acquisition and preprocessing of sensor data and the execution of instructions from the main controller. The digital twin construction and simulation module includes a model management unit, a real-time synchronization unit, and a simulation and deduction unit; The prediction and decision control module includes a stability margin calculation unit, a ground settlement prediction unit, and an optimization decision unit; The execution control module interfaces with the existing outrigger hydraulic control system, receives fine-tuning instructions from the decision control module, and converts them into control current for proportional valves or servo valves to realize outrigger movements.

8. The predictive leveling system for crane outriggers based on digital twins according to claim 7, characterized in that, The digital twin construction and simulation module is deployed on an in-vehicle industrial control computer or a cloud server.

9. The predictive leveling system for crane outriggers based on digital twins according to claim 7, characterized in that, The model management unit is used to store and manage the geometric model, physical model, and dynamic model of the crane; the real-time synchronization unit is used to receive real-time data streams and drive the model to keep synchronized with the physical entity; the simulation derivation unit has a built-in trained reduced-order model and is responsible for performing fast forward simulation.

10. The predictive leveling system for crane outriggers based on digital twins according to claim 7, characterized in that, The stability margin calculation unit calculates the SMI index based on the simulation results; the ground settlement prediction unit runs the ground settlement model to predict the settlement trend. Optimize the predictive control algorithm of the decision-making unit's operating model to solve for the optimal outrigger fine-tuning command.

Images