CN-122014722-A - Multi-source sensing oil cylinder clamping stress non-uniformity prevention control method

CN122014722ACN 122014722 ACN122014722 ACN 122014722ACN-122014722-A

Abstract

The invention relates to the technical field of hydraulic servo cooperative control, in particular to a multi-source sensing oil cylinder clamping stress non-uniformity prevention control method which is applied to a framework comprising a sensing module, a control system and a plurality of groups of servo hydraulic oil cylinders and comprises the steps of collecting displacement, speed, pressure and pipe network flow of each oil cylinder to construct a low-dimensional state sequence, and mapping the low-dimensional state sequence into a high-dimensional state vector by using an observation dictionary matrix; the method comprises the steps of constructing a Coumann linear prediction equation by combining the vector, solving a predicted shear stress vector of a target structural member, calculating a difference value between a safe yield stress threshold value and the predicted stress, constructing a control barrier function, generating a hard constraint affine boundary equation, fusing track deviation and control increment penalty construction cost function, executing convex optimization solution under the hard constraint condition, obtaining an optimal control vector, issuing an oil cylinder to execute, and maintaining a control closed loop. The invention realizes the advanced prejudgment and interception of the stress extreme value and effectively prevents the plastic damage of the component.

Inventors

WANG LIN
LI PING
CAO BIN
PENG MAOLIN

Assignees

深圳市博硕科技股份有限公司

Dates

Publication Date: 20260512
Application Date: 20260323

Claims (10)

1. The multi-source sensing oil cylinder clamping stress non-uniformity prevention control method is applied to a framework comprising a multi-source sensing module, a control system and a plurality of groups of servo hydraulic oil cylinders, and is characterized by comprising the following steps of: S1, acquiring displacement, speed, cavity pressure and total flow of a hydraulic pipe network of each servo hydraulic cylinder based on the multi-source sensing module, and constructing a low-dimensional state sequence; S2, performing dynamic mode decomposition transformation on the low-dimensional state sequence by using an observation dictionary matrix formed by polynomials and radial basis functions, and mapping the low-dimensional state sequence into a high-dimensional state vector; S3, combining the high-dimensional state vector, constructing a Coulomb linear prediction equation by using the identified high-dimensional state transfer matrix and the control input matrix, and obtaining a predicted shear stress vector of the target structural member by observing the mapping matrix; s4, calculating a difference value between a safe yield stress threshold value and the predicted shear stress vector to construct a control barrier function, introducing a convergence coefficient to reconstruct the control barrier function into a recursive inequality, and generating a hard constraint affine boundary equation for limiting a control input vector; S5, constructing a quadratic cost function by combining the track tracking deviation and the control increment penalty, and performing convex optimization solution by taking the hard constraint affine boundary equation as a hard constraint condition to obtain an optimal control vector; and S6, the optimal control vector is issued to each servo hydraulic cylinder to perform driving, and the acquisition operation of the low-dimensional state sequence is triggered again after the action is finished so as to maintain a control closed loop.
2. The method for controlling stress non-uniformity prevention of cylinder clamping by multi-source sensing according to claim 1, wherein in S1, based on the multi-source sensing module collecting displacement, speed, cavity pressure and total flow of a hydraulic pipe network of each servo hydraulic cylinder, constructing a low-dimensional state sequence comprises: synchronously extracting real-time displacement data, real-time speed data, working cavity transient pressure data and transient total flow data of the hydraulic pipe network of each servo hydraulic cylinder according to a set uniform time sampling reference; Cascading and packaging the real-time displacement data, the real-time speed data, the working cavity transient pressure data and the transient total flow data under the same time section according to a set physical dimension sequence to generate a transient state vector corresponding to a single time point; and arranging and combining the transient state vectors of a plurality of continuous acquisition periods according to a time sequence order to construct the low-dimensional state sequence reflecting the dynamic evolution process of the physical system.
3. The method for controlling stress non-uniformity prevention for clamping a cylinder by multi-source sensing according to claim 1, wherein in S2, the mapping the low-dimensional state sequence into a high-dimensional state vector by using an observation dictionary matrix formed by polynomials and radial basis functions comprises: extracting each state variable in the low-dimensional state sequence, and substituting each state variable into a preset polynomial basis function to generate a first mapping feature subset; Substituting each state variable into a radial basis function taking a set state center as a reference to generate a second mapping feature subset; Performing dimension stitching on the first mapping feature subset and the second mapping feature subset to form the observation dictionary matrix; And performing inner product space projection operation on discrete data points in the low-dimensional state sequence by using the observation dictionary matrix, and outputting the high-dimensional state vector with dimensional parameters higher than those of the low-dimensional state sequence.
4. The method for controlling stress non-uniformity prevention of a multi-source-aware cylinder clamping according to claim 1, wherein in S3, combining the high-dimensional state vector, and constructing a kunman linear prediction equation using the identified high-dimensional state transfer matrix and the control input matrix comprises: Invoking the high-dimensional state transfer matrix and the control input matrix which are extracted by least square approximation identification based on a system historical operation data set offline; performing matrix multiplication operation by using the high-dimensional state transfer matrix and the high-dimensional state vector at the current sampling moment to obtain an autonomous evolution component representing the autonomous linear evolution characteristic of the system state; Performing matrix multiplication operation by using the control input matrix and a control input vector at the current sampling moment to obtain an external intervention component representing the intervention characteristic of external control on the system state; And carrying out linear algebraic superposition on the autonomous evolution component and the external intervention component to construct the Coulomb linear prediction equation for predicting the evolution result of the state of the next discrete time step.
5. The method for controlling stress non-uniformity prevention of cylinder clamping by multi-source sensing according to claim 1, wherein in S3, obtaining a predicted shear stress vector of a target structural member by observing a mapping matrix comprises: extracting the observation mapping matrix which is generated in advance based on the calibration of the three-dimensional finite element stiffness model of the target structural member; Extracting a high-dimensional prediction state vector of the next discrete time step deduced and output by the Coumann linear prediction equation; and performing dimension-reducing linear mapping transformation operation on the high-dimensional prediction state vector by using the observation mapping matrix, reversely converting the high-dimensional prediction state vector in Gao Weixi-terbert function space into a physical three-dimensional task space, and extracting the prediction shear stress vector of each stress node of the target structural member at the next discrete time step.
6. The multi-source aware cylinder clamping anti-stress non-uniformity control method according to claim 1, wherein in S4, said calculating a difference between the safe yield stress threshold and the predicted shear stress vector to construct a control barrier function comprises: extracting the safe yield stress threshold value calibrated in advance according to allowable stress limit physical properties of the target structural member material; Setting the safe yield stress threshold value as a system safety upper limit boundary, and performing difference operation on each predicted shear stress component in the predicted shear stress vectors and the safe yield stress threshold value respectively to obtain stress margin vectors corresponding to each stress node; A physical safety set in which the target system does not plastically deform is defined based on non-negative characteristics of each data element in the stress margin vector, and a spatial boundary equation enveloping the physical safety set is defined as the control barrier function at a continuous domain boundary.
7. The method for controlling stress non-uniformity prevention for a cylinder clamping by multi-source sensing according to claim 6, wherein in S4, the introducing a convergence coefficient to reconstruct it into a recursive inequality, generating a hard constraint affine boundary equation for limiting a control input vector comprises: Introducing the convergence coefficient for determining the system state approaching the physical security set boundary evolution rate; constructing the recurrence inequality embodying the forward invariance characteristic by using the obstacle function value at the current moment and the obstacle function value at the next predicting moment and the convergence coefficient; Substituting a control input variable in the Coumann linear prediction equation into the recursive inequality for expansion, separating a linear association term containing the control input variable and a system constant term, reconstructing the linear inequality into a linear inequality aiming at the control input vector, and taking the linear inequality as the hard constraint affine boundary equation.
8. The method for controlling stress non-uniformity prevention of cylinder clamping by multi-source sensing according to claim 1, wherein in S5, the constructing a quadratic cost function by combining trajectory tracking deviation and control increment penalty comprises: Extracting a set reference demolding displacement track sequence, and performing difference comparison on the reference demolding displacement track sequence and an actual displacement state value fed back by a multi-source perception module to generate the track tracking deviation; Extracting control input change rates reflecting the action smoothness of each servo hydraulic oil cylinder, and quantizing the control input change rates by combining a set positive weight matrix to generate the control increment penalty; And carrying out weighted summation operation on the quadratic term of the track tracking deviation and the quadratic term of the control increment penalty, and constructing the quadratic cost function in a mathematical quadratic standard function form.
9. The method for controlling stress non-uniformity prevention of cylinder clamping by multi-source sensing according to claim 1, wherein in S5, performing convex optimization solution on the hard constraint affine boundary equation as a hard constraint condition, and obtaining an optimal control vector comprises: setting the constructed quadratic cost function as a minimum objective function of a convex optimization solving algorithm; setting the hard constraint affine boundary equation as an absolute physical boundary condition which cannot be violated in the optimizing iterative operation process of a convex optimization solving algorithm; And performing iterative search operation in a feasible control input space meeting the absolute physical boundary condition by using an interior point method, calculating a control input solution for enabling the quadratic cost function to obtain the minimum value, and establishing the control input solution as the optimal control vector.
10. The method for controlling stress non-uniformity prevention of cylinder clamping by multi-source sensing according to claim 1, wherein in S6, issuing the optimal control vector to each servo hydraulic cylinder to perform driving, and re-triggering the collection operation of the low-dimensional state sequence after the operation is completed comprises: Converting digital quantity instructions in the optimal control vector into matched analog voltage driving signals; Transmitting the analog voltage driving signals to a servo proportional valve control end in each servo hydraulic cylinder in parallel, and driving each servo hydraulic cylinder to execute associated physical displacement actions according to analog voltage instructions; And generating a trigger interrupt signal at the tail end of a control period when each servo hydraulic cylinder finishes a single driving instruction, waking up a data sampling array of the multi-source sensing module by using the trigger interrupt signal, and forcedly entering a physical parameter extraction and low-dimensional state sequence construction flow of the next discrete time step.

Description

Multi-source sensing oil cylinder clamping stress non-uniformity prevention control method Technical Field The invention relates to the technical field of hydraulic servo cooperative control, in particular to a multisource sensing oil cylinder clamping uneven stress prevention control method. Background In the fields of aerospace and new energy, multiple groups of servo cylinders are required to be scheduled for multidirectional collaborative demolding of complex structural members. Because of strong fluid-solid coupling interference of the pressure fluctuation of the hydraulic pipe network and the asymmetric friction, the existing hysteresis feedback control mechanism extremely depends on the online iteration of the nonlinear jacobian matrix. The method is limited by the computational bottleneck of an industrial controller, and complicated calculus operation causes instruction delay of tens of milliseconds, so that the system cannot predict and intervene the coupling working condition in advance. This results in unbalanced motion of each shaft, and the local nodes accumulate due to mechanical deformation energy so that the shear stress breaks through the yield limit of the material, and finally plastic damage or surface strain of the structural member is initiated. Disclosure of Invention In order to make up for the defects, the invention provides a multisource sensing cylinder clamping uneven stress prevention control method, which aims to solve the problem that components are easy to generate plastic damage because hysteresis feedback is mostly adopted in traditional multi-cylinder control, and force calculation is delayed and stress mutation cannot be prevented in advance. The invention provides a multi-source sensing oil cylinder clamping stress non-uniformity prevention control method, which is applied to a framework comprising a multi-source sensing module, a control system and a plurality of groups of servo hydraulic oil cylinders and comprises the following steps: S1, acquiring displacement, speed, cavity pressure and total flow of a hydraulic pipe network of each servo hydraulic cylinder based on the multi-source sensing module, and constructing a low-dimensional state sequence; S2, performing dynamic mode decomposition transformation on the low-dimensional state sequence by using an observation dictionary matrix formed by polynomials and radial basis functions, and mapping the low-dimensional state sequence into a high-dimensional state vector; S3, combining the high-dimensional state vector, constructing a Coulomb linear prediction equation by using the identified high-dimensional state transfer matrix and the control input matrix, and obtaining a predicted shear stress vector of the target structural member by observing the mapping matrix; s4, calculating a difference value between a safe yield stress threshold value and the predicted shear stress vector to construct a control barrier function, introducing a convergence coefficient to reconstruct the control barrier function into a recursive inequality, and generating a hard constraint affine boundary equation for limiting a control input vector; S5, constructing a quadratic cost function by combining the track tracking deviation and the control increment penalty, and performing convex optimization solution by taking the hard constraint affine boundary equation as a hard constraint condition to obtain an optimal control vector; and S6, the optimal control vector is issued to each servo hydraulic cylinder to perform driving, and the acquisition operation of the low-dimensional state sequence is triggered again after the action is finished so as to maintain a control closed loop. By adopting the technical scheme, a Coulomb prediction equation is constructed, a hard constraint boundary equation is generated by combining a safety yield threshold value to carry out convex optimization solution, so that stress extreme value advanced prediction and interception are realized, and the problem that plastic damage is easy to occur to components due to the fact that hysteresis feedback is mostly adopted in traditional multi-cylinder control, and the stress mutation cannot be prevented in advance due to the fact that calculation force is delayed is solved. Optionally, in S1, based on the multi-source sensing module collecting displacement, speed, cavity pressure and total flow of the hydraulic pipe network of each servo hydraulic cylinder, constructing the low-dimensional state sequence includes: synchronously extracting real-time displacement data, real-time speed data, working cavity transient pressure data and transient total flow data of the hydraulic pipe network of each servo hydraulic cylinder according to a set uniform time sampling reference; Cascading and packaging the real-time displacement data, the real-time speed data, the working cavity transient pressure data and the transient total flow data under the same time section according to a set ph