Physical model and data driving fused hydraulic mechanical arm modeling method and system

Abstract

The invention belongs to the field of hydraulic mechanical arm modeling, and provides a hydraulic mechanical arm modeling method and system fusing a physical model and data driving. The modeling method for the hydraulic mechanical arm fusing the physical model and the data drive comprises the steps that the hydraulic mechanical arm is virtually decomposed into a plurality of mutually independent sub-models, so that a closed chain structure between a joint and a connecting rod is decoupled; constructing a hydraulic driving system model of each sub-model, and further combining the hydraulic driving system model with the corresponding sub-model to form a corresponding subsystem; an embedded neural network model is adopted to learn dynamic characteristics of the hydraulic mechanical arm under different working conditions, and then dynamic compensation is conducted on all subsystems; and combining the subsystems with the dynamic compensation of the embedded neural network models corresponding to the subsystems to construct a dynamic model of the hydraulic mechanical arm. The modeling precision of the hydraulic mechanical arm can be improved, and the response speed of the hydraulic mechanical arm can be increased.

Description

Technical Field

[0001] The present invention belongs to the field of hydraulic mechanical arm modeling, and in particular relates to a hydraulic mechanical arm modeling method and system integrating physical model and data drive. Background Art

[0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art.

[0003] Hydraulic manipulators are automated devices that rely on hydraulic drive. They are composed of multiple connecting rods, joints, and hydraulic actuators, and can achieve complex spatial motion and precision tasks. Compared with electric manipulators, hydraulic manipulators have higher load-bearing capacity and fewer electrical components, and are suitable for heavy loads and harsh operating environments. Hydraulic systems are more suitable for high-risk operating environments such as explosive gases and underwater because they are easy to achieve explosion-proof and waterproof designs. At present, hydraulic robots have become important tools for coal mining, nuclear power maintenance, distribution network detection, and underwater operations, playing an important role in reducing operational risks, improving operational efficiency, reducing costs, and enhancing equipment stability.

[0004] However, the linear hydraulic cylinder of the hydraulic manipulator and the manipulator connecting rod form a closed-chain structure, which leads to a coupling effect between the output force of the linear actuator and the multi-link mechanism. This coupling characteristic makes the precise positioning and motion control of the end of the manipulator challenging. In addition, hydraulic systems often exhibit significant nonlinear characteristics, such as pressure fluctuations and hysteresis effects, which further increase the difficulty of establishing a precise control model for hydraulic manipulators. Traditional modeling and control methods based on motor-driven manipulators are difficult to effectively cope with the complex dynamic characteristics of hydraulic manipulators.

[0005] However, the linear hydraulic cylinder of the hydraulic manipulator forms a closed-chain structure with the manipulator joint, which results in a coupling effect between the output force of the linear actuator and the multi-link mechanism of the arm. This coupling characteristic makes it challenging for the manipulator to perform precise positioning and motion control. In addition, hydraulic systems often exhibit significant nonlinear characteristics, such as pressure fluctuations and hysteresis effects, which further increase the difficulty of developing accurate and stable control models. Traditional modeling and control methods based on motor-driven manipulators are difficult to effectively cope with the complex dynamic characteristics of hydraulic manipulators. Summary of the invention

[0006] In order to solve the above technical problems, the present invention provides a hydraulic robotic arm modeling method and system that integrates physical models and data-driven, which combines a hybrid modeling framework of physics and data-driven, and can establish an accurate dynamic model to improve the accuracy and stability of the hydraulically driven robotic arm in special operating tasks.

[0007] In order to achieve the above object, the present invention adopts the following technical solution:

[0008] A first aspect of the present invention provides a hydraulic mechanical arm modeling method that integrates physical model and data drive.

[0009] In one or more embodiments, a method for modeling a hydraulic mechanical arm integrating a physical model and data-driven is provided, comprising:

[0010] The hydraulic manipulator is virtually decomposed into multiple independent sub-models to decouple the closed chain structure between joints and connecting rods;

[0011] Construct a hydraulic drive system model for each sub-model, and then combine it with the corresponding sub-model to form a corresponding sub-system;

[0012] An embedded neural network model is used to learn the dynamic characteristics of the hydraulic manipulator under different working conditions, and then dynamically compensate each subsystem;

[0013] The dynamic compensation of the subsystems and their corresponding embedded neural network models are combined to construct the dynamic model of the hydraulic manipulator.

[0014] A second aspect of the present invention provides a hydraulic mechanical arm modeling system that integrates physical models and data-driven.

[0015] In one or more embodiments, a hydraulic mechanical arm modeling system integrating physical model and data-driven includes:

[0016] A hydraulic manipulator virtual decomposition module, which is used to virtually decompose the hydraulic manipulator into multiple independent sub-models to decouple the closed chain structure between the joints and the connecting rods;

[0017] A hydraulic drive system model building module is used to build a hydraulic drive system model for each sub-model, and then combine it with the corresponding sub-model to form a corresponding sub-system;

[0018] Subsystem dynamic compensation module, which is used to learn the dynamic characteristics of the hydraulic manipulator under different working conditions using an embedded neural network model, and then dynamically compensate each subsystem;

[0019] The dynamic model building module is used to combine the dynamic compensation of the subsystem and its corresponding embedded neural network model to build a dynamic model of the hydraulic manipulator.

[0020] A third aspect of the present invention provides a computer-readable storage medium.

[0021] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps in the above-mentioned method for modeling a hydraulic mechanical arm integrating a physical model and data-driven modeling.

[0022] A fourth aspect of the present invention provides an electronic device.

[0023] An electronic device comprises a memory, a processor and a computer program stored in the memory and executable on the processor. When the processor executes the program, the steps in the above-mentioned method for modeling a hydraulic mechanical arm integrating a physical model and data-driven are implemented.

[0024] Compared with the prior art, the present invention has the following beneficial effects:

[0025] (1) The present invention designs a physical-data hybrid dynamics model of the robot arm by combining virtual decomposition with an embedded neural network. The robot arm joints are decoupled by virtual decomposition, and the complex mechanical system is divided into multiple simple subsystems. Considering the hydraulic drive model, a complete dynamic model is established for each subsystem, which greatly reduces the complexity of robot arm modeling.

[0026] (2) The physical equation embedded neural network proposed in the present invention combines the learning ability of neural networks and the dynamic characteristics of differential equations. By embedding a physical dynamic system in a neural network and using inverse dynamics equations to construct a specialized neural network to explicitly encode physical knowledge, the network can not only learn the static characteristics of the data, but also capture and simulate the dynamic changes of the data. In a complex operating environment, it effectively ensures that the dynamics of the robot arm are constantly updated and iterated, and the compensation model is continuously updated to adapt to the ever-changing external conditions, thereby improving operating efficiency and safety.

[0027] (3) The physical-data-driven hybrid dynamics model proposed in the present invention is reversible and computationally efficient, and can be applied to inverse dynamics algorithms such as feedforward control strategy and feedback linearization control strategy, effectively improving the control performance of the robot arm operating in complex environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] The accompanying drawings in the specification, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0029] Figure 1 It is a flow chart of a method for modeling a hydraulic mechanical arm integrating a physical model and data-driven according to an embodiment of the present invention;

[0030] Figure 2 is a virtual exploded structural diagram of a swing hydraulic cylinder driving joint according to an embodiment of the present invention;

[0031] Figure 3 is a virtual exploded structural diagram of a linear hydraulic cylinder driven joint according to an embodiment of the present invention;

[0032] Figure 4 is a schematic diagram of an electronic device according to an embodiment of the present invention;

[0033] Figure 5 is a structural diagram of an embedded physical information neural network according to an embodiment of the present invention;

[0034] Figure 6 is a flow chart of physical-data driven hybrid modeling according to an embodiment of the present invention;

[0035] Figure 7 It is a structural schematic diagram of a hydraulic mechanical arm modeling system integrating a physical model and data drive according to an embodiment of the present invention. DETAILED DESCRIPTION

[0036] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0037] It should be noted that the following detailed descriptions are all illustrative and intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present invention belongs.

[0038] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprising" and / or "including" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.

[0039] The prior art provides a virtual decomposition control method for a flexible hydraulic manipulator that takes nonlinear deformation into consideration. The method can realize the decoupling of the closed-chain structure of the hydraulic manipulator joints and realize the virtual decomposition control of multi-module subsystems. The effect of virtual decomposition depends on parameters such as mass, center of mass, and inertia tensor matrix in dynamics. However, most of these parameters are calculated from computer-aided programs, which have obvious deviations from the actual manipulator connecting rod and hydraulic cylinder parameters. Although there are currently adaptive algorithms to adjust these parameters, the accurate estimation and real-time update of these parameters in a dynamically changing environment is still a challenge, and it is difficult to meet the requirements of accurate dynamic modeling for hydraulic manipulators with strong nonlinearity and complex coupling characteristics.

[0040] The prior art also provides a method for adaptive neural network control of hydraulic manipulators, which uses a neural network to approximate the modeling uncertainty of the hydraulic manipulator and dynamically adjusts the system control parameters through an adaptive algorithm. However, the overfitting network in the training data of the manipulator model will result in its inability to effectively generalize to data that has not been collected. More importantly, the generated model lacks physical consistency and interpretability, which makes it complicated to use the results for purposes other than simulation (i.e., control and identification).

[0041] The present invention aims to propose a hybrid modeling framework combining physics and data-driven to solve this problem, and to establish an accurate dynamic model to improve the accuracy and stability of the hydraulically driven robotic arm in special operating tasks.

[0042] The present invention provides a high-precision modeling method for a hydraulic mechanical arm that integrates physical models and data-driven. The present invention uses a virtual segmentation method to achieve dimensionality reduction processing of a multi-degree-of-freedom mechanical arm system. In view of the influence of unknown parameters such as valve dead zone, nonlinear friction, and oil viscosity on system modeling, a neural network that integrates physical information is used to estimate the system state and encode characteristic variables to achieve dynamic nonlinear model compensation. Finally, high-precision modeling of a multi-degree-of-freedom hydraulic mechanical arm is achieved.

[0043] Figure 1 FIG. 1 is a flow chart of a method for modeling a hydraulic mechanical arm that integrates a physical model and data drive in an embodiment of the present invention. Figure 1 The method for modeling a hydraulic mechanical arm by integrating a physical model with data driving in the embodiment shown may include:

[0044] S101, the hydraulic robot arm is virtually decomposed into multiple independent sub-models to decouple the closed chain structure between the joints and the connecting rods.

[0045] In step S101, virtual decomposition theory is used to decouple joints of the hydraulic manipulator, and different types of virtual cutting points are designed for the swing hydraulic cylinder drive joints and the linear hydraulic cylinder drive joints, respectively.

[0046] Among them, a virtual split is added to the rotary joint driven by the swing hydraulic cylinder to decompose it into two independent links, and a coordinate system is added to the end of each link close to the base of the robot arm, such as Figure 2 shown.

[0047] Coordinate system A i1 The generalized velocity at can be obtained by forward iteration of the velocity of the previous part of the link of the virtual split, then the driving joint A i2 The generalized velocity at is:

[0048]

[0049] zτi =[000001] T

[0050] In the formula is the angular velocity of active joint i, Represents coordinate system A i2 is relative to the coordinate system A i1 The transformation matrix is ​​used to transform the coordinate system A i1 The generalized force / velocity vector in is transformed into coordinate system A i2 The equivalent force vector in z τi The bit is a mapping matrix that maps joint angular velocity to generalized velocity.

[0051] Generalized speed and Substituting into the formula, we can get the force required for each member, and further get the internal force relationship of all members of the structure as follows:

[0052]

[0053] in Represents in coordinate system A i The force received by the lower link is Represents the force on the end of the connecting rod, the matrix Represents the transformation matrix of coordinate system T relative to coordinate system B.

[0054] The required output force of the hydraulic system is: where f ci is the output force of the hydraulic cylinder corresponding to the i-th joint.

[0055] For the variable triangle closed chain structure driven by the linear hydraulic cylinder, two virtual divisions are used, such as Figure 3 As shown, the closed chain structure is first cut into an open chain, and then virtual cutting points are added to the connection between the hydraulic cylinder and the joint, and between the cylinder and the piston rod, respectively, to obtain four relatively independent connecting rods.

[0056] In the closed chain structure, the relationship between the joint angle displacement and the linear hydraulic cylinder displacement is as follows:

[0057]

[0058] Among them l j1 and l j2 is the length of two adjacent links of the robot arm, x j0 is the effective length of the hydraulic linear actuator at zero piston stroke, q j is the angle value of the jth joint of the robot arm, q j0 is the initial angle value of the jth joint of the robot, α j , β jis the intrinsic value of the jth joint of the robot.

[0059] The generalized velocity vector in the jth hydraulic joint can be written as follows:

[0060] V Aj11 =V Aj

[0061]

[0062] Where: mapping matrix z j1 =[000001] T 、z j2 =[100000] T ; V Aj Represents in coordinate system A j The generalized velocity vector under ; is the connecting rod displacement speed.

[0063] The generalized force relationship in the jth hydraulic joint is as follows:

[0064] F Tj =F Tj1 +F Tj2

[0065] F Tj1 =α1F Tj +η j

[0066] F Tj2 =α2F j -η j

[0067] Where: η j is the internal force between the open chain j1 and the open chain j2; α1, α2 are proportional distribution factors, and α1+α2=1; where F Tj Represents the generalized force received by the end of the connecting rod, F Tj1 Indicates the force on the end of the robotic arm, F Tj2 Indicates the force on the end of the robotic arm.

[0068] In this embodiment, there are 3 passive joints and 2 active joints in the closed chain structure, and the output force required by the hydraulic cylinder is: f cj =z j F j2 .

[0069] S102, constructing a hydraulic drive system model for each sub-model, and then combining it with the corresponding sub-model to form a corresponding sub-system.

[0070] The flow equation for the hydraulic servo valve is:

[0071]

[0072] Where: k qi is relative to the valve core displacement x vi The flow gain; p 1i , Q 1i and p 2i , Q 2i are the pressure and flow of the rodless chamber and the rod chamber of the hydraulic cylinder respectively; p si is the oil supply pressure, p ti is the oil return pressure; s(x vi ) is the defined selection function.

[0073] During operation, the characteristics of the hydraulic oil remain unchanged, and the pressure in each chamber of the linear hydraulic cylinder is equal. The dynamic equation of pressure-flow in the two chambers can be described as:

[0074]

[0075] Where: β i is the elastic modulus of the oil; C i is the leakage coefficient; is the displacement speed of the hydraulic cylinder; V 1i 、V 2i Respectively represent the volume of the rod chamber and the volume of the rodless chamber of the hydraulic linear cylinder; A 1i , A 2i is the piston area of ​​the rod chamber and the piston area of ​​the rodless chamber of the hydraulic linear cylinder

[0076] Substituting the servo valve flow equation into the physical model of the hydraulic system can be expressed by the pressure of the two chambers:

[0077]

[0078] Where: k is the extension and retraction speed of the hydraulic cylinder; vi is the servo valve voltage gain, x vi =k vi u i ; The volume of the two chambers of the hydraulic cylinder is V 1i =V 01i +A 1i x i , V 2i =V 02i -A 2i x i .

[0079] The linear hydraulic cylinder output driving force f of the i-th joint of the robot arm i for:

[0080] f i =A 1i p1i -A 2i p 2i

[0081] Linear hydraulic cylinder output force f i The relationship with the output torque is:

[0082]

[0083] Among them J j is the length of the linear hydraulic cylinder x j Joint rotation q j The differential of

[0084] The swing hydraulic cylinder of the i-th joint of the robot arm has an output torque τ i Related to displacement as follows:

[0085]

[0086] τ i =V i p 1i -V i p 2i

[0087] Where: V i is the displacement of the i-th joint swing hydraulic cylinder; τ i is the output torque of the i-th joint; p 1i 、p 2i is the pressure in the two chambers of the swing lever.

[0088] S103, using an embedded neural network model to learn the dynamic characteristics of the hydraulic manipulator under different working conditions, and then dynamically compensate each subsystem.

[0089] In order to compensate for the unknown friction force and system modeling errors existing in the physical model, this embodiment uses an embedded neural network model to dynamically compensate each subsystem.

[0090]

[0091] in is a vector composed of a set of nonlinear basis functions, which is a known physical nonlinear unit; j , p 1j ,p 2j They represent the joint position variable, velocity variable, and cylinder pressure variable of the jth joint collected, respectively. j Functions represent a series of known mathematical relationships, consisting of squares, radicals, various types of quadratic operations, etc., which are used as custom layers to simulate the basic mathematical combinations of physical nonlinear units; β jis the neuron weight coefficient in the neural network; ρ j As a hidden bias term, it corresponds to the nonlinear friction of the system.

[0092] Linear friction torque τ fj The following friction model is used:

[0093]

[0094] where τ fj represents the friction torque of the jth joint; f vj is the viscous friction coefficient of the jth joint; f cj is the Coulomb friction coefficient of the jth joint; τ offj Contains the additional offset of the jth joint caused by the asymmetric Coulomb friction coefficient and sensor noise.

[0095] The specific form of the neural network compensation model can be obtained, such as Figure 5 As shown, its expression is:

[0096]

[0097] Where P represents the number of joint parameters; β p Represents the weight coefficient between nonlinear mapping functions; h jp () represents the overall nonlinear mapping function; Respectively represent the joint variables q after nonlinear mathematical transformation j The corresponding weight coefficients in the polynomial; Represents the joint variable q j The weight coefficient of Represents the joint velocity variable q j The weight coefficient of and Indicates the weight coefficient corresponding to the hydraulic cylinder pressure; β p+1 Indicates the viscous friction coefficient in the hydraulic system; β p+2 represents the Coulomb friction coefficient; β p+3 represents the position noise coefficient; the subscript j represents the j-th joint.

[0098] In order to improve the performance of the neural network, a custom loss function is designed that takes into account both the neuron parameters and the physical weights of the system, as shown in the following formula:

[0099]

[0100] Among them, loss is the loss function of the embedded neural network model; n k is the total number of joint points; τ j is the true torque of the jth joint; τ j ′ is the predicted torque of the jth joint; βinj is the corresponding polynomial weight coefficient; is a physical nonlinear function. The first half is the mean square error between the predicted torque and the actual torque, and the second half is the physical formula related to the robot arm.

[0101] S104, combining the dynamic compensation of the subsystem and its corresponding embedded neural network model to construct a dynamic model of the hydraulic robot arm.

[0102] For a given multi-DOF manipulator model, the dynamic equation of the link represented by the jth joint in its fixed branch coordinate system Aj (i.e., Newton-Euler equation) can be written as follows:

[0103]

[0104] where j∈{1,…,n}, M j ,C j ,G j Denote the branched chain coordinate system A j Inertia matrix, Coriolis force matrix and gravity matrix, represents the torque required by the robot arm, ω j ,V j represent the linear velocity and angular velocity vectors respectively.

[0105] The torque required for the j-th joint of the robot The hydraulic drive torque τ j and friction torque τ fj and unknown modeling errors. According to the proposed method, the dynamic model of the multi-DOF manipulator can be written as follows. Its specific structure is as follows: Figure 6 As shown:

[0106]

[0107] in In branched coordinate system A j Compensation torque in

[0108] Reference Figure 4 , a schematic diagram of an electronic device is given. It should be noted that, Figure 4 The electronic device 400 shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0109] like Figure 4As shown, electronic device 400 includes a central processing unit (CPU) 401, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 402 or a program loaded from a storage portion 408 into a random access memory (RAM) 403. In RAM 403, various programs and data required for system operation are also stored. Central processing unit 401, ROM 402 and RAM 403 are connected to each other via bus 404. Input / output (I / O) interface 405 is also connected to bus 404.

[0110] The following components are connected to the I / O interface 405: an input section 406 including a keyboard, a mouse, etc.; an output section 407 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 408 including a hard disk, etc.; and a communication section 409 including a network interface card such as a local area network (LAN) card, a modem, etc. The communication section 409 performs communication processing via a network such as the Internet. A drive 410 is also connected to the I / O interface 405 as needed. A removable medium 411, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 410 as needed so that a computer program read therefrom is installed into the storage section 408 as needed.

[0111] When the central processing unit 401 in the electronic device of this embodiment executes the program, the following is achieved: Figure 1 The steps in the method of integrating physical model and data-driven hydraulic manipulator modeling are shown.

[0112] In particular, according to an embodiment of the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present application includes a computer program product, which includes a computer program carried on a computer readable medium, the computer program including a computer program for executing Figure 1 In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 409, and / or installed from the removable medium 411. When the computer program is executed by the central processing unit 401, various functions defined in the apparatus of the present application are executed.

[0113] in, Figure 1 The computer program instructions corresponding to the method shown may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce a product including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0114] Figure 7 is a schematic diagram of a hydraulic mechanical arm modeling system that integrates physical model and data drive in an embodiment of the present invention. Figure 1 The fusion physical model of the hydraulic manipulator corresponds to the data-driven modeling method, such as Figure 7 As shown, the hydraulic mechanical arm modeling system integrating physical model and data-driven in this embodiment may include:

[0115] A hydraulic mechanical arm virtual decomposition module 701, which is used to virtually decompose the hydraulic mechanical arm into multiple independent sub-models to decouple the closed chain structure between the joints and the connecting rods;

[0116] A hydraulic drive system model building module 702 is used to build a hydraulic drive system model for each sub-model, and then combine it with the corresponding sub-model to form a corresponding sub-system;

[0117] A subsystem dynamic compensation module 703, which is used to use an embedded neural network model to learn the dynamic characteristics of the hydraulic mechanical arm under different working conditions, and then dynamically compensate each subsystem;

[0118] The dynamic model building module 704 is used to combine the dynamic compensation of the subsystem and its corresponding embedded neural network model to build a dynamic model of the hydraulic mechanical arm.

[0119] It should be noted here that Figure 7 The fusion physical model in the data-driven hydraulic manipulator modeling system is combined with the various modules in the data-driven hydraulic manipulator modeling system. Figure 1 The fusion physical model in corresponds one to one with each step in the data-driven hydraulic robotic arm modeling method, and the specific implementation process is the same, which will not be repeated here.

[0120] A person skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program, and the program can be stored in a computer-readable storage medium, and when the program is executed, it can include the processes of the embodiments of the above-mentioned methods. The storage medium can be a disk, an optical disk, a read-only memory (ROM) or a random access memory (RAM), etc.

[0121] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for modeling a hydraulic mechanical arm integrating physical model and data drive, characterized in that: include: The hydraulic manipulator is virtually decomposed into multiple independent sub-models to decouple the closed chain structure between joints and connecting rods; Construct a hydraulic drive system model for each sub-model, and then combine it with the corresponding sub-model to form a corresponding sub-system; An embedded neural network model is used to learn the dynamic characteristics of the hydraulic manipulator under different working conditions, and then dynamically compensate each subsystem; The dynamic compensation of the subsystems and their corresponding embedded neural network models are combined to construct the dynamic model of the hydraulic manipulator.

2. The method for modeling a hydraulic mechanical arm integrating a physical model and data drive according to claim 1, characterized in that: Virtual decomposition theory is used to decouple the joints of the hydraulic robot arm, and different types of virtual cutting points are designed for the swing hydraulic cylinder driven joints and the linear hydraulic cylinder driven joints respectively.

3. The method for modeling a hydraulic mechanical arm integrating a physical model and data drive as claimed in claim 2, characterized in that: Add a virtual split to the rotary joint driven by the swing hydraulic cylinder to decompose it into two independent links, and add a coordinate system at the end of each link close to the base of the robot arm.

4. The method for modeling a hydraulic mechanical arm integrating a physical model and data drive as claimed in claim 2, characterized in that: For the variable triangular closed chain structure driven by the linear hydraulic cylinder, two virtual segmentations are used. First, the closed chain structure is cut into an open chain. Then, virtual cutting points are added to the connection between the hydraulic cylinder and the joint, and between the cylinder and the piston rod, respectively, to obtain four relatively independent connecting rods.

5. The method for modeling a hydraulic mechanical arm integrating a physical model and data drive according to claim 1, characterized in that: The loss function of the embedded neural network model is: Among them, loss is the loss function of the embedded neural network model; n k is the total number of joint points; τ j is the true torque of the jth joint; τ j ′ is the predicted torque of the jth joint; β inj is the corresponding polynomial weight coefficient; is a physical nonlinear function, They respectively represent the joint position variable, velocity variable, and cylinder pressure variable of the j-th joint collected.

6. The method for modeling a hydraulic mechanical arm integrating a physical model and data drive according to claim 1, characterized in that: The expression of the dynamic model of the hydraulic manipulator is: where j∈{1,…,n}, M j ,C j ,G j Denote the branched chain coordinate system A j Inertia matrix, Coriolis force matrix and gravity matrix, ω j ,V j Denote the linear velocity and angular velocity vectors, τ j In the branched coordinate system A j The driving torque of the manipulator in j In the branched coordinate system A j The compensation torque in .

7. The method for modeling a hydraulic mechanical arm integrating a physical model and data drive as claimed in claim 6, characterized in that: In the branched coordinate system A j The expression of the compensation torque in is: Where P represents the number of joint parameters; β p Represents the weight coefficient between nonlinear mapping functions; h jp () represents the overall nonlinear mapping function; Respectively represent the joint variables q after nonlinear mathematical transformation j The corresponding weight coefficients in the polynomial; Represents the joint variable q j The weight coefficient of Represents the joint velocity variable q j The weight coefficient of and Indicates the weight coefficient corresponding to the hydraulic cylinder pressure; β p+1 Indicates the viscous friction coefficient in the hydraulic system; β p+2 represents the Coulomb friction coefficient; β p+3 Represents the position noise factor.

8. A hydraulic mechanical arm modeling system integrating physical model and data drive, characterized in that: include: A hydraulic manipulator virtual decomposition module, which is used to virtually decompose the hydraulic manipulator into multiple independent sub-models to decouple the closed chain structure between the joints and the connecting rods; A hydraulic drive system model building module is used to build a hydraulic drive system model for each sub-model, and then combine it with the corresponding sub-model to form a corresponding sub-system; Subsystem dynamic compensation module, which is used to learn the dynamic characteristics of the hydraulic manipulator under different working conditions using an embedded neural network model, and then dynamically compensate each subsystem; The dynamic model building module is used to combine the dynamic compensation of the subsystem and its corresponding embedded neural network model to build a dynamic model of the hydraulic manipulator.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps in the method for modeling a hydraulic mechanical arm that integrates a physical model and data drive as described in any one of claims 1 to 7 are implemented.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the steps in the method for modeling a hydraulic mechanical arm that integrates a physical model and data drive as described in any one of claims 1 to 7 are implemented.

Images