Method for determining pushing stroke of tube rolling machine based on multi-action pose cooperation and tube rolling machine

Abstract

The application relates to the field of underground pipeline construction equipment, and particularly discloses a method for determining a pushing stroke of a pipe rolling machine based on multi-action pose cooperation and the pipe rolling machine. The method comprises the following steps: a multi-actuator-stratum coupling physical model is established, and an initial pushing stroke path is obtained through inversion; multi-action pose data are collected in real time during construction, actual response vectors are constructed, and dynamic error vectors are formed by comparing the actual response vectors with model prediction values; a poor response path section is identified along a time axis, adaptive disturbance is applied to control parameters of the poor section to generate a new generation path; a response weight matrix is introduced to weight error components, and the best pushing stroke path is output and closed-loop executed until convergence is achieved by taking the weighted error norm as a criterion and iteratively optimizing until convergence is achieved. Through deep integration of parameter inversion before construction and multi-generation path iteration during construction, the application realizes adaptive dynamic optimization of the pushing stroke, and significantly improves multi-cylinder cooperation accuracy and whole-machine verticality control capability.

Description

Technical Field

[0001] This invention belongs to the field of underground pipeline construction equipment, specifically a method for determining the pushing stroke of a pipe rolling machine based on multi-action posture coordination, and a pipe rolling machine. Background Technology

[0002] A pipe rolling machine is a specialized construction device used for underground pipeline laying. Its working principle involves the coordinated action of multiple hydraulic actuators (including a propulsion cylinder, lifting cylinder, clamping cylinder, and adjusting cylinder) to press steel casing sections into the ground. During construction, the propulsion cylinder provides axial thrust to press the casing down; the clamping cylinder holds the casing and, in conjunction with a rotary drive, causes slight twisting of the casing to reduce sidewall friction; the lifting and adjusting cylinders are used to adjust the overall verticality of the machine, ensuring the casing is driven into the ground in the designed direction. Because there is a complex kinematic coupling relationship between the displacement of each hydraulic cylinder and the overall machine posture, determining the stroke parameters (i.e., the displacement-time sequence of each hydraulic cylinder) is the core element for achieving precise multi-cylinder coordination.

[0003] However, in actual construction, existing pipe rolling machines often rely on operator experience to set the pushing stroke parameters or use fixed preset values, failing to dynamically adjust them based on real-time feedback information such as displacement, thrust, and verticality deviation collected during construction. When stratum conditions change, the preset pushing stroke parameters deviate from the actual situation, leading to inconsistent strokes of various actuators, excessive verticality deviation of the entire machine, and consequently, construction problems such as machine tilting, single-cylinder overload, and even pipe jamming. Therefore, how to adaptively optimize the pushing stroke path based on real-time feedback during construction is a pressing technical challenge in this field. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention provides a method for determining the push stroke of a pipe rolling machine based on multi-action posture coordination, as well as a pipe rolling machine itself. This solves the problem that the push stroke parameters in the prior art cannot be dynamically and adaptively optimized based on real-time construction feedback, resulting in poor multi-cylinder coordination accuracy and difficulty in ensuring the verticality of the entire machine.

[0005] This invention provides a method for determining the pushing stroke of a tube rolling machine based on multi-action pose coordination, comprising the following steps:

[0006] Step S1: Pre-construction modeling and parameter inversion

[0007] Establish a multi-actuator-stratum coupling physical model;

[0008] With the objective of minimizing the error between the model's predicted response and the reference construction response, the initial push path is obtained through inversion. ;

[0009] Step S2: Dynamic Feedback and Multi-Generation Path Iteration during Construction

[0010] S2.1: Deploy sensor groups at each actuator and the machine body of the tube rolling machine to collect multi-motion pose data in real time and construct the actual response vector. ;

[0011] S2.2: Combine the actual response vector with the path derived from the current propagation. The response vector predicted by the multi-actuator-formation coupling physical model Comparisons are made to form a dynamic error vector. ;

[0012] S2.3: The time axis along the push stroke will Segmentation is used to identify inferior response path segments whose Euclidean norm of the error vector exceeds a preset threshold, denoted as... ;

[0013] S2.4: Apply a disturbance to the control parameters corresponding to the inferior response path segment to generate a new generation of push path. ;

[0014] S2.5: Using a feedback weight matrix The weighted error vector is obtained by weighting the components of the dynamic error vector. The weighted error vector Euclidean norm is calculated based on each control cycle. In continuous The moving average value over each control period As a basis for judgment;

[0015] like And at the same time In continuous If the condition is met within each control cycle, convergence is determined, and the current generation path is output as the optimal propagation path. ;

[0016] Otherwise, repeat steps S2.2 to S2.5 until the above convergence condition is met;

[0017] in, This is the preset allowed value; This is the preset convergence threshold; It is a preset positive integer;

[0018] Step S3: Closed-loop execution

[0019] The optimal push path The displacement-time command sequence of each hydraulic cylinder is converted into a sequence of commands and sent to the corresponding actuator driver to drive the pipe rolling machine to complete the casing pressing operation. At the same time, the real-time data generated during construction is sent back to step S2 to form an adaptive closed-loop control.

[0020] Preferably, in step S1, the multiple actuators in the model include at least a propulsion cylinder, a lifting cylinder, a clamping cylinder, and an adjusting cylinder, and the number of actuators is assumed to be... The model predicts the response, including the displacement, velocity, thrust of each actuator, and the verticality deviation of the entire machine.

[0021] The inversion is constrained by physical limitations: the upper limit of stroke, maximum thrust, and synchronicity tolerance of each hydraulic cylinder. The objective function to minimize the response error is:

[0022] ;

[0023] , , ;

[0024] in, Indicates the first Predicted displacement value of each cylinder Compared with reference value difference, Indicates the first The maximum stroke of each cylinder Predicted value representing verticality deviation Compared with reference value difference, Indicates the maximum permissible verticality deviation. Indicates the first Predicted thrust per cylinder Compared with reference value difference, Indicates the first The maximum permissible thrust of each cylinder , , This represents the weighting coefficient, used to balance target items with different dimensions and different importance.

[0025] Preferably, in step S2, the method for identifying poor response path segments is as follows:

[0026] Calculate the average error vector within each time window. and its variance ;

[0027] when and When this happens, mark the window as a poor response path segment;

[0028] in, and The sample mean and variance of the error vector for the initial undisturbed phase are given. The initial undisturbed phase refers to the construction phase within the first 10 seconds after the first casing section begins to be pressed down. , The values ​​are preset positive constants, and their ranges are respectively: , .

[0029] Preferably, in step S2.4, the perturbation function used to generate the next-generation push path is... Defined as:

[0030] ;

[0031] in, For the disturbance amplitude coefficient, , The attenuation coefficient is... , It is a symbolic function; It represents the Hadamah accumulation. Describes the Euclidean norm. This represents the dynamic error vector, used to guide subsequent path corrections; the sign of the error determines the direction of adjustment. Indicates the current number Proxy travel route This indicates a poorly performing response path segment.

[0032] Preferably, the feedback weight matrix It is a diagonal matrix, and its diagonal elements According to the Historical accumulation of actuator errors and dynamic adjustment of construction priorities:

[0033] ;

[0034] in, For the first The implementing agencies recently The average of the absolute values ​​of the errors at each sampling time. It is a positive integer. For learning rate, ; Indicates all The maximum value among the absolute values ​​of the average error of each actuator; and for the updated Normalization is performed to make ,in The number of implementing agencies.

[0035] Preferably, the sensor group includes: a wire-type displacement sensor installed on the propulsion cylinder and the lifting cylinder, a pressure sensor installed on the clamping cylinder, a magnetostrictive displacement sensor installed on the adjusting cylinder, and a dual-axis inclinometer installed on the body of the pipe rolling machine; the posture data includes at least the real-time stroke of each cylinder, clamping force, propulsion force, overall machine levelness and verticality.

[0036] Preferably, the multi-actuator-stratum coupling physical model includes the following sub-models:

[0037] Hydraulic system dynamics sub-model: describes the flow-pressure-displacement relationship of each cylinder;

[0038] Formation resistance sub-model: Calculate axial friction and end resistance based on casing penetration depth and geological exploration data;

[0039] Pose coupling sub-model: The mapping relationship between the displacement of each actuator and the pose of the whole machine is established by using the kinematic equations of multiple rigid bodies;

[0040] The model's predicted response It is obtained by solving a set of differential-algebraic equations through numerical integration.

[0041] The present invention also provides a tube rolling machine employing the above method, comprising:

[0042] Organism;

[0043] Multiple hydraulic actuators, including at least one propulsion cylinder, at least two lifting cylinders, at least one clamping cylinder and at least one adjusting cylinder, are used to clamp the casing, adjust its verticality and push it downward.

[0044] Sensor arrays are arranged at various actuators and the body to collect pose data in real time;

[0045] Controller, the controller includes:

[0046] Parameter inversion module: used to establish a multi-actuator-stratum coupled physical model and invert the initial push path;

[0047] Dynamic error calculation module: used to compare measured data with model predictions and generate dynamic error vectors;

[0048] The path iteration optimization module is used to identify poorly responding path segments, apply perturbations, and iteratively generate the best push path.

[0049] Closed-loop execution module: used to convert the optimal push stroke path into drive instructions and issue them to each actuator;

[0050] A communication bus connects the sensor group, controller, and hydraulic actuator driver.

[0051] Preferably, the controller further includes a pose cooperative solver, which calculates the current overall verticality deviation based on the position. Based on the stroke margin of each cylinder, the collaborative allocation coefficient is calculated in real time, and the stroke command of each cylinder is adjusted according to the following rules:

[0052] The safety threshold is defined as 10% of the maximum stroke of each actuator;

[0053] like Then, the stroke command of the actuator closer to the tilted side is increased, and the stroke command of the actuator on the opposite side is decreased. The preset verticality deviation threshold has a range of values. ;

[0054] If the stroke margin of all actuators is lower than the safety threshold, the stroke redistribution mode is triggered: the single propulsion stroke of the propulsion cylinder is reduced, and the lifting cylinder and adjusting cylinder share part of the verticality compensation task to avoid single cylinder overload.

[0055] Compared with the prior art, the present invention has the following beneficial effects:

[0056] By establishing a multi-actuator-stratum coupled physical model, the initial push path is obtained by inversion before construction with the goal of minimizing the error between the model's predicted response and the reference construction response, so that the initial parameters have the best predictive ability for the stratum conditions.

[0057] During construction, sensors are used to collect multi-motion pose data in real time. The measured response is compared with the model prediction response to form a dynamic error vector. Poor response path segments are identified along the time axis, and adaptive perturbations are applied to the control parameters of these poor segments based on the error sign and amplitude to generate a new generation of push stroke paths. A feedback weight matrix is ​​introduced to differentiate the weighting of the error components of each actuator, using the weighted error norm as the convergence criterion. Through multiple generations of path iterations, the optimal push stroke path is continuously approximated. Finally, the optimal path is transformed into a hydraulic cylinder displacement-time command sequence for execution, while construction data is simultaneously fed back to form a closed loop.

[0058] This method achieves a deep integration of pre-construction parameter inversion and multi-generation iterative optimization during construction, enabling the pipe rolling machine to adapt to changes in the strata, significantly improving the multi-cylinder coordination accuracy and overall verticality control capability, effectively avoiding problems such as machine tilting, single-cylinder overload or pipe jamming caused by inconsistent stroke, and improving construction quality and system robustness. Attached Figure Description

[0059] Figure 1 This is a flowchart of the method for determining the push stroke in a tube rolling machine based on multi-action pose coordination in Embodiment 1 of the present invention;

[0060] Figure 2 This is a schematic diagram of a tube rolling machine.

[0061] In the diagram: 1. Body; 2. Propulsion cylinder; 3. Lifting cylinder; 4. Clamping cylinder; 5. Adjustment cylinder. Detailed Implementation

[0062] The technical solutions of 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.

[0063] Example 1: As Figure 1 As shown, this invention provides a method for determining the pushing stroke in a pipe rolling machine based on multi-action posture coordination. The pushing stroke refers to the coordinated displacement planning during the pushing phase, that is, the displacement-time coordinated scheduling scheme of each hydraulic actuator during the pushing construction phase of the pipe rolling machine (i.e., during the casing pressing operation). The method includes the following steps:

[0064] Step S1: Pre-construction modeling and parameter inversion

[0065] 1. Construction of a multi-actuator-stratum coupled physical model

[0066] Establish the following sub-model:

[0067] a. Hydraulic system dynamics sub-model

[0068] The flow-pressure-displacement relationship of each hydraulic cylinder is described by the following differential equation:

[0069] ;

[0070] ;

[0071] in, For the inflow of the first The flow rate of each hydraulic cylinder (controlled by an electro-hydraulic proportional valve). For the effective area of ​​the piston, For displacement, Leakage coefficient, For hydraulic pressure, This refers to the thrust output by the hydraulic cylinder.

[0072] It should be noted that the inflow of the first Flow rate of each hydraulic cylinder It is determined based on the flow conservation equation, and is composed of the sum of the "effective volume increment" used to drive the piston movement and the "ineffective volume flow" lost due to leakage, specifically:

[0073] The piston motion causes the working chamber volume to change at a rate of 100%. For a single-rod cylinder, the area of ​​the rod-side chamber is used when the rod is extended, and the area of ​​the rodless chamber is used when the rod is retracted. The effective area is set as follows: ;

[0074] The leakage rate is assumed to be laminar leakage, which is proportional to the pressure difference, i.e. The pressure difference is the difference between the working pressure and the return oil pressure, and the return oil pressure is approximately 0.

[0075] Therefore, the inflow rate is the sum of the effective flow rate and the leakage flow rate. .

[0076] b. Formation resistance sub-model

[0077] Total axial resistance encountered during casing insertion End resistance Friction with sidewalls composition:

[0078] ;

[0079] in, express Total axial resistance from the formation to the casing at any given time (unit: The direction is vertically upward, opposite to the downward pressing direction of the propulsion cylinder. This indicates the resistance experienced at the bottom (end) of the casing, originating from the end-bearing force exerted when the casing bottom cuts / compacts the soil. This represents the frictional resistance between the casing sidewall and the surrounding soil, distributed along the entire length of the casing.

[0080] Based on geological exploration data, the strata are divided into several layers along the depth, and each layer is assigned a friction coefficient. and end drag coefficient When the depth of the bottom of the casing Located in the Layer time:

[0081] ;

[0082] ;

[0083] in, The outer diameter of the casing. For the first The thickness of the layer, For the front The cumulative thickness of the layers, The area of ​​the bottom end of the casing. Indicates the first Sidewall friction coefficient of the layer (unit: ), Indicates the first Sidewall friction coefficient of the layer (unit: This reflects the frictional strength between the casing sidewall and the soil layer (for cohesive soil, the undrained shear strength is commonly used for calculation; for sandy soil, the effective internal friction angle is commonly calculated using the lateral pressure coefficient). Indicates the first End drag coefficient of layer (unit: This reflects the end bearing resistance per unit area of ​​the soil layer on the bottom of the casing. This indicates the outer diameter circumference of the casing (unit: m). Indicates the casing at the first The sidewall area (cylindrical side area) of the layer. Indicates the bottom end of the casing is at the current [number]. The depth of entry into a layer (i.e., the depth already entered within the current layer). This represents the frictional resistance between the casing sidewall and the soil in the current layer (only the side area of ​​the already entered portion is considered).

[0084] c. Pose Coupler Submodel

[0085] The tube rolling machine is simplified into a multi-rigid-body system, and its kinematic equations are established.

[0086] Define the body coordinate system displacement of the propulsion cylinder , and lifting cylinder displacement , perpendicularity deviation of the whole machine The relationship is:

[0087] ;

[0088] in, The distance between the left and right lifting cylinders. The distance between the left and right propulsion cylinders. , These represent the displacements (in meters) of the left and right propulsion cylinders, respectively. , These represent the displacements of the left and right lifting cylinders, respectively (unit: m). This indicates the verticality deviation of the entire machine (unit: radians, converted to degrees below), set as the angle between the longitudinal (or vertical) direction of the machine body and the direction of gravity, with a target value of 0 (perfectly vertical).

[0089] By combining the above sub-models, a set of differential algebraic equations (DAE) is formed. The fourth-order Runge-Kutta method is used for numerical integration to predict the path of a given push. System response This includes the displacement, velocity, thrust, and verticality deviation of each cylinder, among which... This indicates the displacement of the clamping cylinder, used to control the clamping force to hold the sleeve tightly. This indicates the displacement of the adjusting cylinder, used to assist in fine-tuning the verticality.

[0090] 2. Inversion of the initial travel path

[0091] The goal of inversion is to find a set of initial displacement assignment parameters. (in (corresponding to six actuators), so that the error between the model's predicted response and the reference construction response is minimized;

[0092] It should be noted that before actual construction begins, the "reference construction response" refers to the expected response set based on historical similar strata data and conventional construction experience. This expected response reflects the quality indicators of qualified construction.

[0093] Using a weighted least squares objective function:

[0094] ;

[0095] , , ;

[0096] in, Indicates the first Predicted displacement value of each cylinder Compared with reference value The difference (unit: m) Indicates the first Maximum stroke per cylinder (unit: m). Predicted value representing verticality deviation Compared with reference value difference, Indicates the maximum permissible verticality deviation. Indicates the first Predicted thrust per cylinder Compared with reference value difference, Indicates the first The maximum permissible thrust of each cylinder, for the propulsion cylinder is... Clamping cylinder is , , , This represents the weighting coefficient, used to balance target items with different dimensions and importance. In this embodiment, we take... , , It emphasizes the importance of displacement and perpendicularity; the maximum allowable perpendicularity deviation is... .

[0097] Physical constraints include:

[0098] Upper limit of stroke of each cylinder: .

[0099] Maximum thrust constraint: (propulsion cylinder) Clamping cylinder ).

[0100] Synchronization tolerance: the difference in displacement between the left and right propulsion cylinders at any given time. ; Displacement difference between left and right lifting cylinders .

[0101] The constrained nonlinear optimization problem described above is solved using the Particle Swarm Optimization (PSO) algorithm with a population size of 50 and 100 iterations to obtain the initial push path. and with The time interval is discretized into an instruction sequence.

[0102] Step S2: Dynamic Feedback and Multi-Generation Path Iteration during Construction

[0103] S2.1. Real-time data acquisition and actual response vector construction

[0104] After construction begins, the controller operates in each control cycle. Internal sensor sample values ​​( (That is, averaging 10 sampling points per period) to construct the actual response vector:

[0105] ;

[0106] Among them, speed The thrust is obtained from the displacement center difference. Converted from pressure sensor, This represents the measured displacement of the six hydraulic cylinders. This represents the measured speed of the six hydraulic cylinders. This represents the measured thrust of the six hydraulic cylinders. This indicates the measured verticality deviation of the entire machine, directly measured by a biaxial inclinometer. This represents the actual response vector constructed.

[0107] S2.2. Calculation of Dynamic Error Vector

[0108] In the current number Proxy travel path Below, a multi-actuator-formation coupled physical model is used to predict the response. Then calculate the dynamic error vector:

[0109] ;

[0110] in, Indicates the iteration path sequence number ( This is the initial path. (The path after the first update) Indicates the first The system response predicted by the model under the alternative path. Represents the dynamic error vector (time step). The difference between the measured and predicted values ​​is used to guide subsequent path corrections, and the sign of the error determines the direction of adjustment.

[0111] S2.3. Identification of Inferior Response Path Segments

[0112] The entire push stroke timeline Divided into lengths of A continuous time window; for each window Calculate the mean of the error vector within the window (considering only the displacement components). and variance .

[0113] Average reference error during normal construction phase and variance The data was obtained offline from the first 10 seconds of casing insertion (during this stage, the formation is usually homogeneous and the resistance is stable); empirical coefficients were taken as... , .

[0114] If both conditions are met:

[0115] ;

[0116] Then mark the window as a poor response path segment. .

[0117] S2.4. Adaptive Perturbation and Next-Generation Path Generation

[0118] For time intervals marked as inferior response path segments, a disturbance is applied to the control parameters (i.e., the displacement commands of each hydraulic cylinder) within that time period; the disturbance function is defined as follows:

[0119] ;

[0120] in:

[0121] Let be the disturbance amplitude coefficient, taken as... (That is, the initial disturbance amplitude is 10% of the current displacement command).

[0122] Let be the attenuation coefficient, and take . .

[0123] The sign vector is obtained by taking the sign of each component of the error vector.

[0124] This represents the Hadamard product (element-by-element multiplication).

[0125] Design principle: This perturbation function has the characteristic of "small perturbation for large errors, and large perturbation for small errors"; when the error norm When it is large, the exponent term As the error norm approaches zero, the disturbance amplitude automatically decays, preventing system oscillations or divergence caused by excessively large correction steps; as the error norm gradually decreases, the exponential term approaches 1, and the disturbance amplitude recovers to its normal value. At this point, a fine search is performed to approximate the optimal solution; this conservative-aggressive two-stage strategy ensures the stability and convergence of the iterative process.

[0126] Then, a new generation of push path is generated:

[0127] ;

[0128] For periods not marked as poor responses, the perturbation is 0, meaning the original path is maintained.

[0129] S2.5. Weighted Error and Convergence Judgment

[0130] Introducing a feedback weight matrix It is a diagonal matrix, with diagonal elements According to the The historical average error of each actuator is dynamically adjusted; its update formula is:

[0131] ;

[0132] in, For the first The implementing agencies recently Each sampling time (corresponding to) The absolute value of the average displacement error, and the learning rate. , This represents the maximum absolute value of the average error among all 6 actuators, used for normalization;

[0133] After the update, all ownership is re-normalized, making Initially all .

[0134] The weighted error vector is calculated as follows:

[0135] ;

[0136] The Euclidean norm of the weighted error vector is used as a convergence criterion.

[0137] ;

[0138] The convergence condition is: the average weighted error norm over the entire push stroke time axis is less than a threshold. And the absolute value of the overall verticality deviation at each moment All are less than the allowable value The former ensures that the weighted sum of displacement deviations of each cylinder is minimized, while the latter ensures that the verticality meets the construction quality requirements.

[0139] If the convergence condition is met, stop the iteration and output the current path as the optimal push path. Otherwise, Repeat steps S2.2 to S2.5.

[0140] In this embodiment, convergence is typically achieved after 3 to 5 iterations. During the iteration process, the actuators with large error contributions are adjusted first (their weights are automatically increased) to quickly eliminate coordination inconsistencies.

[0141] Step S3: Closed-loop execution

[0142] The controller will push the optimal travel path. Converted into current commands for each electro-hydraulic proportional valve, to The data is periodically sent to each hydraulic cylinder driver, while sensor data is continuously collected and sent back to step S2. If the geological conditions suddenly change during construction (such as encountering isolated rocks), the dynamic error vector norm will exceed the threshold again, and the system will automatically trigger a new round of multi-generation path iteration to achieve adaptive closed-loop control.

[0143] Specific rules for the pose co-solver:

[0144] The controller also integrates a pose cooperative solver for real-time fine-tuning during iteration intervals (i.e., during single-generation path execution) to further improve cooperative accuracy. This solver operates independently of the iterative optimization process in step S2 and at a higher frequency (per control cycle) as a supplement to multi-generation path iteration.

[0145] Define a high threshold for verticality deviation Safety threshold (That is, the travel margin is less than 10% of the maximum travel).

[0146] Rule 1: When At that time, the solver calculates the cooperative allocation coefficient. :

[0147] ;

[0148] in The distance between the two actuators. The proportional coefficient is used, and the lateral deviation is the sign of the current tilt direction of the aircraft relative to the direction perpendicular to the target. Then, the stroke command of the actuator closer to the tilt side is increased first, while the stroke command of the actuator on the opposite side is decreased, with the range of command change limited to... Each control cycle.

[0149] Rule 2: When the travel margin of all actuators is less than When this occurs, the "trip reassignment" mode is triggered:

[0150] Reduce the single propulsion stroke command of the propulsion cylinder by 30%;

[0151] The lifting cylinder and adjusting cylinder share the verticality compensation task originally undertaken by the propulsion cylinder, based on the verticality deviation. Specifically, this is achieved by temporarily adjusting the target value of the verticality deviation to... To achieve this.

[0152] This mode effectively prevents a single cylinder from seizing or overloading due to exhaustion of its stroke.

[0153] Self-learning database and remote monitoring:

[0154] The controller integrates a self-learning database that stores the following information for each construction operation: geological type, depth segmentation, and the final push path used. The parameters include the mean error norm and the algebra required for convergence. When similar geological conditions are encountered again, the parameter inversion module directly retrieves the historical best solution from the database as the initial value. This can reduce the number of iterations by more than 50%.

[0155] Remote monitoring and early warning unit The frequency of the dynamic error vector norm change rate is calculated. If the rate of change exceeds 5 seconds for 5 consecutive seconds If the iterative optimization fails to converge within 3 generations, it is determined to be a "sudden change in the formation", an audible and visual warning signal is issued, and the propulsion speed is automatically reduced to 50% of the original speed, waiting for confirmation from the operator.

[0156] Example 2: As Figure 2 As shown, this embodiment provides a pipe rolling machine, including: a machine body 1, a push cylinder 2, a lifting cylinder 3, a clamping cylinder 4, an adjusting cylinder 5, a rotary drive device, and a control system.

[0157] Propulsion cylinder 2: There are two of them, used to provide axial thrust for pressing down the casing, with a maximum stroke. Maximum thrust .

[0158] Lifting cylinder 3: There are two of them, used to adjust the levelness and verticality of the machine body 1, with a maximum stroke. .

[0159] Clamping cylinder 4: One cylinder is used to clamp the sleeve, with a maximum clamping force. .

[0160] Adjusting cylinder 5: Quantity is one, used for fine adjustment of verticality, maximum stroke .

[0161] The sensor group includes:

[0162] Wire-type displacement sensor (range) precision (This is a system installed at the piston rod ends of the propulsion cylinder 2 and the lifting cylinder 3, used to collect the displacement of each cylinder in real time.) .

[0163] Pressure sensor (range) precision It is installed in the oil circuit of clamping cylinder 4 to collect clamping force. .

[0164] Magnetostrictive displacement sensor (range) precision ), installed at the end of the piston rod of the regulating cylinder 5.

[0165] Biaxial Inclinometer (Measuring Range) precision It is installed on the base plane of the machine body 1 and is used to measure the levelness of the entire machine in real time. and verticality ;

[0166] Define verticality deviation Target verticality .

[0167] All sensors The sampling frequency transmits data to the controller via the CAN bus; the controller in each control cycle ( ,Right now The average value of data from multiple sampling points within a given period is taken as the actual response value for that period.

[0168] Furthermore, to verify the beneficial effects of the adaptive iterative adjustment method based on multi-action pose coordination and dynamic error driving proposed in this invention, a comparative experiment was designed. Five comparative methods were selected, and a comprehensive evaluation was conducted from multiple dimensions, including push stroke displacement tracking accuracy, overall verticality control stability, convergence speed, and anti-interference timeliness, to verify the superiority of the proposed solution, as detailed below:

[0169] Experimental environment: This experiment was conducted on an industrial control computer equipped with an Intel Core i7-12700K processor and 32GB DDR4 memory. The software environment was the MATLAB R2022a simulation platform, which was used in conjunction with AMESim 2020.1 for co-simulation of hydraulic system dynamics.

[0170] The physical model parameters of the pipe rolling machine are based on the equipment specifications (lifting cylinder stroke) given in Example 2. Clamping force of clamping cylinder The control cycle is set to 0.2m (adjusting cylinder stroke). Sensor sampling frequency .

[0171] Experimental data: The experimental data comes from 82 sets of valid construction data collected at a certain engineering site, combined with simulation data under different geological conditions generated by MIDAS GTSNX finite element software, and includes a total of Group of samples (each group of samples corresponds to one section of the cannula (length)) The complete pressing process, duration ),according to The proportion is divided into training set ( The group is used for historical database initialization and parameter calibration, and the test set ( The group is used for performance evaluation of each method, with three typical working conditions: uniform soft soil, alternating soft and hard soil, and composite strata containing boulders.

[0172] Five comparison methods are set up, specifically:

[0173] Method 1 (Solution of this invention): The method proposed in this invention is an adaptive iterative adjustment method based on multi-action pose coordination and dynamic error driving, which includes four core components: multi-actuator-stratum coupled physical modeling and initial path inversion, identification of inferior response path segments based on time window and adaptive perturbation, dynamic adjustment of feedback weight matrix and multi-generation path iteration, and real-time fine-tuning of pose coordination solver.

[0174] Method 2 (compared to Method 1): This method employs a traditional fixed-parameter open-loop control method. Before construction, a set of fixed thrust parameters is set based on the geological exploration report (the displacement of each cylinder is proportionally and synchronously extended, and the thrust speed is constant). During construction, it does not rely on sensor feedback at all, nor does it have any closed-loop adjustment mechanism. It relies entirely on the operator's experience for manual intervention.

[0175] Method 3 (Compared to Method 2): Using a proportional-integral approach based on a single verticality deviation. The feedback adjustment method uses only the overall verticality deviation measured by a biaxial inclinometer. As the sole feedback signal, a uniform force is applied to the four propulsion / lifting cylinders. Correction amount ( , 05), but without establishing a formation resistance model, and without decoupling and optimizing the independent displacement errors of each cylinder, there is a problem of mutual coupling of the adjustment commands of each cylinder.

[0176] Method 4 (compared to Method 3): An offline batch adjustment method based on mean statistics is adopted. This method adjusts the batch size in each construction section (each advance). Approximately corresponding After completion, calculate the average value of each sensor data within that segment, compare it with the expected value, and then manually adjust the push stroke parameters for the next segment. The adjustment interval is approximately... It lacks real-time dynamic response capabilities, and the correction cycle is on the order of minutes.

[0177] Method 5 (Ablation Experiment Method): The ablation version of the present invention is adopted, that is, the "Identification of inferior response path segments and adaptive perturbation based on time window" module is removed (the initial modeling inversion in step S1 and the weighted error convergence judgment in step S2.5 are retained, but the dynamic perturbation generation and path update mechanism based on time window in S2.3 and S2.4 is deleted and replaced with a fixed step size). (Uniform correction) to verify the contribution of the adaptive perturbation strategy to the overall convergence performance.

[0178] Four evaluation indicators are set, including: push stroke displacement tracking accuracy, root mean square deviation of overall machine verticality, convergence speed, and anti-interference robustness coefficient, specifically:

[0179] The stroke displacement tracking accuracy, used to measure the weighted average relative deviation between the actual displacement of the six hydraulic cylinders and the target displacement command, reflects the accuracy of the method for coordinated control of multiple actuators. Therefore:

[0180] ;

[0181] in, , , For the measured displacement, For the expected instruction, For the maximum stroke of each cylinder, This indicates the accuracy of displacement tracking during the push stroke; the lower the value, the higher the tracking accuracy.

[0182] The root mean square deviation of the overall verticality is used to measure the degree of fluctuation in the overall verticality deviation of the machine throughout the construction process, reflecting the method's ability to control the verticality quality of the hole. Therefore:

[0183] ;

[0184] in, This represents the perpendicularity deviation measured by a biaxial inclinometer. This represents the root mean square deviation of the overall verticality; the lower the value, the more stable the verticality control.

[0185] Convergence rate, used to measure the equivalent convergence time required for a method to first meet the convergence condition from the start of construction, is as follows:

[0186] ;

[0187] For methods that lack iterative characteristics (methods two, three, and four), if the convergence condition cannot be met throughout the entire process, then take... And marked "not converged". This indicates the convergence speed; a lower value indicates faster convergence.

[0188] The robustness coefficient against disturbances measures the average settling time required for the method to reconverge to a steady state when encountering sudden geological changes (such as isolated boulders). It reflects the method's ability to withstand disturbances. Specifically:

[0189] ;

[0190] in, The number of formation mutations set for each working condition (respectively in...) , , When an instantaneous resistance step is applied, the amplitude is equal to that of the normal resistance. times), The time required for the error norm to fall back to within 90% of the pre-mutation level after a mutation occurs. This represents the robustness coefficient against interference; the lower the value, the stronger the anti-interference capability.

[0191] Performance comparisons were conducted based on computational evaluation metrics, as shown in the table below:

[0192] Table 1: Performance Comparison Results of Various Methods

[0193] method Push stroke displacement tracking accuracy (%) Root mean square deviation of overall machine verticality (°) Convergence speed (s) Anti-interference robustness coefficient (s) Method 1 0.87 0.042 28.5 8.5 Method 2 9.54 0.318 120 Unable to converge autonomously Method 3 6.23 0.176 65.2 28.3 Method 4 4.91 0.135 45.0 35.7 Method 5 2.18 0.089 42.3 18.6

[0194] It is not difficult to see from the comparison table above:

[0195] Compared with method two, the accuracy of the push stroke displacement tracking of the present invention is improved by approximately (from Down to The root mean square deviation of the overall verticality was reduced by approximately (from Down to The convergence speed decreased from failing to converge throughout the entire process. The convergence condition was never met (the time limit was significantly reduced to only a few seconds). (about (iteration)

[0196] Compared with method three, the root mean square deviation of the overall verticality of the present invention is reduced by approximately (from Down to The convergence speed is from shortened to The increase reached ;

[0197] Compared with method four, the present invention outperforms method four in terms of push-stroke displacement tracking accuracy and anti-interference robustness coefficient. from Down to )and (from Down to );

[0198] Analysis of Method 5 reveals that, comparing the experimental results of Method 1 (the complete solution of this invention) and Method 5 (ablation version), removing the "time window-based inferior response path segment identification and adaptive perturbation" decreases the accuracy of push-stroke displacement tracking by approximately [percentage missing]. (from Deteriorated to The convergence speed is from Increase to (Added approximately) The robustness coefficient against interference is from Rise to .

[0199] Table 2: Performance and stability verification results of the present invention under different geological complexities.

[0200] Scene Push stroke displacement tracking accuracy (%) Root mean square deviation of overall machine verticality (°) Convergence speed (s) Anti-interference robustness coefficient (s) Scene 1 0.52 0.028 18.2 5.2 Scene 2 0.91 0.045 29.7 8.9 Scene 3 1.36 0.067 42.1 14.7 average value 0.93 0.047 30.0 9.6

[0201] It is not difficult to see from the comparison table above:

[0202] The present invention achieves a displacement tracking accuracy fluctuation range of only [value missing] under three different conditions: uniform soft soil layer, alternating soft and hard strata, and complex composite strata. (relative to the mean) The fluctuation range is approximately The root mean square deviation of perpendicularity shall not exceed (far below the allowable value) The convergence speeds in the three scenarios are as follows: , and Both are significantly faster than fixed open-loop control. (non-convergence) and PI regulation ( );

[0203] From scenario 1 to scenario 3, although the geological complexity increases significantly (sidewall friction coefficient from...), leap to (and introduces isolated rock impact), the indicators of the present invention still show a gradual rather than abrupt trend, and the standard deviation of each indicator remains within a reasonable range.

[0204] Therefore, the adaptive iterative adjustment method based on multi-action pose coordination and dynamic error driving proposed in this invention shows significant performance advantages in multiple dimensions such as push stroke displacement tracking accuracy, overall verticality deviation control, convergence speed, and anti-interference robustness. It effectively solves the problems in the prior art, such as fixed open-loop control being unable to cope with formation changes, single-signal PI adjustment leading to coordination mismatch, and offline batch correction response lag and blind correction. It improves the automated construction quality and safety of pipe rolling machines under complex formation conditions and verifies the beneficial effects of the technical solution of this invention.

[0205] The embodiments of the present invention are given for the purposes of illustration and description. Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Any changes, modifications, substitutions and variations made by those skilled in the art to the above embodiments within the scope of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for determining the pushing stroke in a tube rolling machine based on multi-action pose coordination, characterized in that, Includes the following steps: Step S1: Pre-construction modeling and parameter inversion Establish a multi-actuator-stratum coupling physical model; With the objective of minimizing the error between the model's predicted response and the reference construction response, the initial push path is obtained through inversion. ; Step S2: Dynamic Feedback and Multi-Generation Path Iteration during Construction S2.1: Deploy sensor groups at each actuator and the machine body of the tube rolling machine to collect multi-motion pose data in real time and construct the actual response vector. ; S2.2: Combine the actual response vector with the path derived from the current propagation. The response vector predicted by the multi-actuator-formation coupling physical model Comparisons are made to form a dynamic error vector. ; S2.3: The time axis along the push stroke will Segmentation is used to identify inferior response path segments whose Euclidean norm of the error vector exceeds a preset threshold, denoted as... ; S2.4: Apply a disturbance to the control parameters corresponding to the inferior response path segment to generate a new generation of push path. ; S2.5: Using a feedback weight matrix The weighted error vector is obtained by weighting the components of the dynamic error vector. The weighted error vector Euclidean norm is calculated based on each control cycle. In continuous The moving average value over each control period As a basis for judgment; like And at the same time In continuous If the condition is met within each control cycle, convergence is determined, and the current generation path is output as the optimal propagation path. ; Otherwise, repeat steps S2.2 to S2.5 until the above convergence condition is met; in, This is the preset allowed value; This is the preset convergence threshold; It is a preset positive integer; Step S3: Closed-loop execution The optimal push path The displacement-time command sequence of each hydraulic cylinder is converted into a sequence of commands and sent to the corresponding actuator driver to drive the pipe rolling machine to complete the casing pressing operation; at the same time, the real-time data generated during construction is sent back to step S2 to form an adaptive closed-loop control. In step S2.4, the perturbation function used to generate the next generation of push path is... Defined as: ； in, For the disturbance amplitude coefficient, , The attenuation coefficient is... , It is a symbolic function; It represents the Hadamah accumulation. Denotes the Euclidean norm. This represents the dynamic error vector, used to guide subsequent path corrections; the sign of the error determines the direction of adjustment. Indicates the current number Proxy travel route This indicates a poorly performing response path segment.

2. The method according to claim 1, characterized in that, In step S1, the multiple actuators in the model include at least a propulsion cylinder, a lifting cylinder, a clamping cylinder, and an adjusting cylinder. Let the number of actuators be... The model predicts the response, including the displacement, velocity, thrust of each actuator, and the verticality deviation of the entire machine. The inversion is constrained by physical limitations: the upper limit of stroke, maximum thrust, and synchronicity tolerance of each hydraulic cylinder. The objective function to minimize the response error is: ； ， ， ； in, Indicates the first Predicted displacement value of each cylinder Compared with reference value difference, Indicates the first The maximum stroke of each cylinder Predicted value representing verticality deviation Compared with reference value difference, Indicates the maximum permissible verticality deviation. Indicates the first Predicted thrust per cylinder Compared with reference value difference, Indicates the first The maximum permissible thrust of each cylinder , , This represents the weighting coefficient, used to balance target items with different dimensions and different importance.

3. The method according to claim 1, characterized in that, In step S2, the method for identifying poor response path segments is as follows: Calculate the average error vector within each time window. and its variance ; when and When this happens, mark the window as a poor response path segment; in, and The sample mean and variance of the error vector for the initial undisturbed phase are given. The initial undisturbed phase refers to the construction phase within the first 10 seconds after the first casing section begins to be pressed down. , The values ​​are preset positive constants, and their ranges are respectively: , .

4. The method according to claim 1, characterized in that, The feedback weight matrix It is a diagonal matrix, and its diagonal elements According to the Historical accumulation of actuator errors and dynamic adjustment of construction priorities: ； in, For the first The implementing agencies recently The average of the absolute values ​​of the errors at each sampling time. It is a positive integer. For learning rate, ; Indicates all The maximum value among the absolute values ​​of the average error of each actuator; and for the updated Normalization is performed to make ,in The number of implementing agencies.

5. The method according to claim 1, characterized in that, The sensor group includes: a wire-type displacement sensor installed on the propulsion cylinder and the lifting cylinder, a pressure sensor installed on the clamping cylinder, a magnetostrictive displacement sensor installed on the adjusting cylinder, and a dual-axis inclinometer installed on the body of the pipe rolling machine; the posture data includes at least the real-time stroke of each cylinder, clamping force, propulsion force, overall machine levelness and verticality.

6. The method according to claim 1, characterized in that, The multi-actuator-stratum coupling physical model includes the following sub-models: Hydraulic system dynamics sub-model: describes the flow-pressure-displacement relationship of each cylinder; Formation resistance sub-model: Calculate axial friction and end resistance based on casing penetration depth and geological exploration data; Pose coupling sub-model: The mapping relationship between the displacement of each actuator and the pose of the whole machine is established by using the kinematic equations of multiple rigid bodies; The model's predicted response It is obtained by solving a set of differential-algebraic equations through numerical integration.

7. A tube rolling machine employing the method described in claim 1, characterized in that, include: Organism; Multiple hydraulic actuators, including at least one propulsion cylinder, at least two lifting cylinders, at least one clamping cylinder and at least one adjusting cylinder, are used to clamp the casing, adjust its verticality and push it downward. Sensor arrays are located at various actuators and the body to collect pose data in real time; Controller, the controller includes: Parameter inversion module: used to establish a multi-actuator-stratum coupled physical model and invert the initial push path; Dynamic error calculation module: used to compare measured data with model predictions and generate dynamic error vectors; The path iteration optimization module is used to identify poorly responding path segments, apply perturbations, and iteratively generate the best push path. Closed-loop execution module: used to convert the optimal push stroke path into drive instructions and issue them to each actuator; A communication bus connects the sensor group, controller, and hydraulic actuator driver.

8. The tube rolling machine according to claim 7, characterized in that, The controller also includes a pose co-solver, which calculates the current overall verticality deviation based on the position. Based on the stroke margin of each cylinder, the coordination distribution coefficient is calculated in real time, and the stroke command of each cylinder is adjusted according to the following rules: The safety threshold is defined as 10% of the maximum stroke of each actuator; like Then, the stroke command of the actuator closer to the tilted side is increased, and the stroke command of the actuator on the opposite side is decreased. The preset verticality deviation threshold has a range of values. ; If the stroke margin of all actuators is lower than the safety threshold, the stroke redistribution mode is triggered: the single propulsion stroke of the propulsion cylinder is reduced, and the lifting cylinder and adjusting cylinder share part of the verticality compensation task to avoid single cylinder overload.

Images