A pneumatic artificial muscle force-controlled actuator for polishing inner surface of deep cavity and a control method thereof

CN122807910APending Publication Date: 2026-09-25FUZHOU ZHONGAO TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611180701.7
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-05
Publication Date
2026-09-25

AI Technical Summary

Technical Problem

然而,气动人工肌肉在抛光应用中仍面临若干技术问题:一是气动人工肌肉本身输出轴向拉力,而抛光作业需要在工具端形成稳定法向作用力;二是气动人工肌肉存在明显的非线性、时变性和迟滞特性,建模和控制难度较高;三是现有部分气动人工肌肉抛光结构仍需要借助连杆机构完成力传递转换,导致径向尺寸增加,不利于狭窄空间进入

Benefits of technology

[0033]1. 本发明采用固定式半圆柱壳体与移动式半圆柱壳体形成沿执行器轴向延伸的细长圆柱外形和内部两侧包覆式圆柱形腔体,整体结构细长紧凑,径向尺寸小,能够进入孔、槽及深腔等狭窄空间实施内表面抛光。

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122807910A_ABST
    Figure CN122807910A_ABST
Patent Text Reader

Abstract

The present application relates to a kind of pneumatic artificial muscle force control executor for deep cavity inner surface polishing and its control method.The fixed half-cylindrical shell of executor and movable half-cylindrical shell jointly form the elongated cylindrical shape along the axial extension of executor and the two sides of internal wrapped cylindrical cavity, pneumatic artificial muscle is arranged in the cylindrical cavity, and with intermediate connecting piece, tension-compression force sensor, movable half-cylindrical shell and polishing mechanism constitute axial series direct transmission type force transmission chain, to directly transmit the axial tension of pneumatic artificial muscle to polishing mechanism without setting internal connecting rod conversion mechanism, so that the normal output force of polishing tool is formed to the inner surface of workpiece.The present application is compact in structure, small in radial dimension, high in rigidity, suitable for entering hole, slot and deep cavity and other narrow space, and can be effectively applied to constant force polishing operation of inner surface of workpiece.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of robotic polishing technology, and in particular to a robotic end effector based on pneumatic artificial muscles for polishing the inner surfaces of holes, grooves and deep cavities, and its working method. Background Technology

[0002] Polishing is a crucial process in the machinery manufacturing industry for improving workpiece surface roughness, enhancing surface quality, and ensuring consistent machining. With the widespread application of industrial robots in automated manufacturing, utilizing robots to replace manual labor in continuous contact tasks such as polishing, grinding, and deburring has become an important development trend.

[0003] In robotic polishing, the end effector needs to maintain stable contact with the workpiece surface, and the normal contact force during this contact process directly affects the workpiece surface quality, material removal uniformity, and processing stability. Therefore, how to achieve constant control of the end effector's output contact force has become a key issue in robotic polishing technology.

[0004] Existing robotic force-controlled polishing actuators mostly employ servo motors, voice coil motors, cylinders, or pneumatic-electric hybrid drive solutions. These actuators have achieved good results in polishing the outer surface of workpieces, but their overall radial dimensions are usually large, and some structures rely on internal linkages, push rods, or complex transmission mechanisms, making it difficult to enter narrow spaces such as holes, grooves, and deep cavities for inner surface polishing.

[0005] For workpieces with holes, grooves, cup-shaped cavities, and deep cavities, their inner surfaces often require processing such as burr removal, tool mark repair, roughness improvement, and flatness enhancement. Due to space constraints and the size limitations of existing end effectors, polishing of these areas usually still relies mainly on manual labor, resulting in high labor intensity, poor processing consistency, low efficiency, and difficulty in maintaining stable quality control.

[0006] Pneumatic artificial muscles, as flexible actuation components, possess advantages such as compact structure, light weight, high power density, and good compliance, making them particularly suitable for force control operations in space-constrained environments. However, pneumatic artificial muscles still face several technical challenges in polishing applications: First, the pneumatic artificial muscle itself outputs axial tensile force, while polishing operations require a stable normal force to be generated at the tool end; second, pneumatic artificial muscles exhibit significant nonlinearity, time-varying characteristics, and hysteresis, making modeling and control difficult; third, some existing pneumatic artificial muscle polishing structures still require linkage mechanisms to complete force transmission and conversion, resulting in an increase in radial dimensions, which is detrimental to entry into narrow spaces.

[0007] Therefore, there is an urgent need to propose a robot end effector with force control polishing, which is compact in structure, has small radial dimensions, does not require an internal linkage conversion mechanism, can be used for polishing the inner surfaces of holes, grooves and deep cavities, and can achieve stable constant force control, as well as its working method. Summary of the Invention

[0008] The purpose of this invention is to overcome the above-mentioned shortcomings of the prior art and provide a robotic end-effector force-controlled polishing actuator based on pneumatic artificial muscles for polishing the inner surfaces of holes, grooves and deep cavities, and its working method; the actuator has a compact structure and forms a stable normal output force at the polishing tool end; the working method is conducive to achieving high-precision control of polishing contact force.

[0009] To achieve the above objectives, the present invention adopts the following technical solution:

[0010] A robotic force-controlled actuator for deep cavity internal polishing includes a flange mounting plate, a fixed semi-cylindrical housing, a movable semi-cylindrical housing, a force control mechanism, a polishing mechanism, a guiding mechanism, an air source interface, an electro-proportional valve, and a controller.

[0011] The flange mounting plate is located at one end of the fixed semi-cylindrical housing near the industrial robot and is used to connect to the end flange of the industrial robot.

[0012] The fixed semi-cylindrical housing and the movable semi-cylindrical housing are arranged opposite each other along the actuator axis, and the two cover each other in the circumferential direction to form a slender cylindrical shape extending along the actuator axis and an internal cylindrical cavity covering both sides. The movable semi-cylindrical housing only moves in a linear axial direction relative to the fixed semi-cylindrical housing under the restriction of the guide mechanism.

[0013] The force control mechanism is disposed in the cylindrical cavity and includes a pneumatic artificial muscle, an intermediate connector, a tension / compression sensor, a displacement sensor, and two sets of spring guide assemblies. One end of the force control mechanism is fixed to the bottom of the fixed semi-cylindrical housing, and the other end is connected to one end of the tension / compression sensor via the intermediate connector. The other end of the tension / compression sensor is connected to the movable semi-cylindrical housing. The displacement sensor is arranged parallel to the pneumatic artificial muscle, and the two sets of spring guide assemblies are symmetrically arranged on both sides of the pneumatic artificial muscle.

[0014] The polishing mechanism is located at the end of the movable semi-cylindrical housing away from the tension / compression sensor, and moves synchronously with the movable semi-cylindrical housing along the actuator axis;

[0015] The air source interface is connected to an external air source, and the electric proportional valve is connected in series between the air source interface and the pneumatic artificial muscle.

[0016] The controller is electrically connected to the tension / compression sensor, the displacement sensor, and the electro-proportional valve, respectively.

[0017] The pneumatic artificial muscle, intermediate connector, tension / compression sensor, movable semi-cylindrical housing, and polishing mechanism are arranged in sequence along the actuator axis to form an axial series direct transmission chain. The actuator does not have a linkage conversion mechanism for changing the direction of force transmission. The axial tension generated by the contraction of the pneumatic artificial muscle is directly transmitted to the polishing mechanism through the axial series direct transmission chain, so that the polishing tool forms a normal output force on the inner surface of the workpiece.

[0018] Preferably, the spring guide assembly includes a guide shaft, a linear bearing, a bearing fixing end, and a compression spring. The bearing fixing end is fixed to the inner wall of the fixed semi-cylindrical housing, and the linear bearing is mounted on the bearing fixing end. One end of the guide shaft is fixedly connected to the intermediate connecting member, and the other end passes through the linear bearing and can slide freely along the axial direction. The compression spring is sleeved on the outside of the guide shaft, and its two ends abut against the end face of the linear bearing and the side face of the intermediate connecting member, respectively. The compression spring has a preset pre-compression amount in the initial assembly state of the actuator to balance the gravity of the moving side component in the non-working state of the actuator and to buffer output force fluctuations in the working state.

[0019] Preferably, the guiding mechanism includes at least three sets of linear guide slider assemblies; each set of linear guide slider assemblies is evenly distributed along the circumference of the actuator, and the linear guide rail of the linear guide slider assembly is axially fixed to the outer wall of the fixed semi-cylindrical housing, and the slider of the linear guide slider assembly is correspondingly fixed to the inner wall of the movable semi-cylindrical housing, which is used to improve the overall rigidity of the actuator and limit the torsion and sway during the polishing process.

[0020] Preferably, the polishing mechanism includes a drive motor, a coupling, and a polishing tool; the drive motor is fixedly mounted on a movable semi-cylindrical housing, and the output shaft of the drive motor is coaxially connected to the clamping end of the polishing tool through the coupling, for driving the polishing tool to rotate around its own axis; the rotation axis of the polishing tool coincides with the axial center line of the actuator, ensuring that the polishing contact force is uniformly output along the normal direction.

[0021] The present invention also provides a method for operating the above-mentioned actuator, comprising the following steps:

[0022] Step A1: Fix the force control actuator to the end effector of the industrial robot using the flange mounting plate, and power on both the force control actuator and the industrial robot; input the preset initial air pressure into the pneumatic artificial muscle through the electro-proportional valve to keep the actuator in its initial state; control the industrial robot to move the actuator to the polishing initial point, so that the working end of the polishing tool maintains a preset safe distance from the surface of the workpiece to be polished without contact.

[0023] Step A2: Input the target contact force value, preset polishing trajectory parameters and controller initial parameters into the controller; start the drive motor of the polishing mechanism to drive the polishing tool to rotate, and at the same time control the electric proportional valve to supply air to the pneumatic artificial muscle, so that the actuator enters the constant force closed-loop control mode;

[0024] Step A3: The controller collects the contact force signal F(k) output by the tension and compression sensors and the contraction displacement signal x(k) output by the displacement sensor in real time at a fixed sampling period; the displacement change, velocity and acceleration at the current sampling moment are calculated based on the contraction displacement signal x(k), which together constitute the current system state variables; combined with the initial parameters of the actuator dynamics model obtained offline, a model input vector for online estimation of model parameters and rolling prediction is constructed.

[0025] Step A4: Based on the three-element dynamic model including the contraction element, elastic element and damping element, establish the discrete prediction model of the actuator; within the set rolling prediction time domain, based on the current system state variables and the model input vector, predict the state response and contact force output response of multiple future sampling steps;

[0026] Step A5: Calculate the deviation between the predicted contact force response and the actual collected contact force signal. Combined with the current system state observations, use the recursive least squares algorithm to estimate and update the parameters of the discrete prediction model online and in real time. Based on the updated discrete prediction model, with the target contact force as the tracking target, and combined with the constraints of air pressure output limit, pneumatic artificial muscle stroke limit, and contact force change rate limit, solve the optimal control air pressure sequence in a rolling manner, and take the first term of the optimal control sequence as the control air pressure p(k) at the current sampling time.

[0027] Step A6: The controller converts the control air pressure p(k) into a corresponding control signal and outputs it to the electro-proportional valve to adjust the input air pressure of the pneumatic artificial muscle in real time. During the process of the industrial robot driving the actuator to move along the inner surface to be polished, the normal contact force between the polishing tool and the inner surface of the workpiece is kept stable at the target value through closed-loop control, so as to complete the constant force polishing operation of the inner surface of the deep cavity.

[0028] Preferably, in step A3, the controller amplifies and converts the analog signal from the tension / compression sensor to digital signal using a force transmitter, and acquires the digital displacement of the displacement sensor through a displacement acquisition circuit; the contraction displacement signal x(k) is first processed by a first-order low-pass filter to remove noise, and then the velocity and acceleration are calculated through differential operation to improve the accuracy of the state quantity.

[0029] Preferably, in step A4, the controller first determines the preset pressure range to which the input air pressure of the current pneumatic artificial muscle (7) belongs, and matches the corresponding equivalent stiffness parameter according to the pressure range; then compares the contraction displacement value of the current sampling period with that of the previous sampling period to determine whether the actuator is currently in the contraction stroke or the extension stroke, and matches the corresponding equivalent damping parameter according to the stroke direction; and substitutes the matched equivalent stiffness and equivalent damping parameters into the three-element dynamic model to update the parameter matrix of the discrete prediction model in order to adapt to the nonlinear and hysteresis characteristics of the pneumatic artificial muscle.

[0030] Preferably, in step A6, the surface adaptive adjustment logic of the actuator when moving along the inner surface of the curved surface with the robot is as follows: when the surface to be polished is concave and the distance between the robot end flange and the surface to be polished increases, the contraction rate of the pneumatic artificial muscle increases, the output tension decreases, and the contact force is lower than the target value. The controller controls the electro-proportional valve to increase the output air pressure, so that the contact force quickly rises back to the target range. When the surface to be polished is convex and the distance between the robot end flange and the surface to be polished decreases, the contraction rate of the pneumatic artificial muscle decreases, the output tension increases, and the contact force is higher than the target value. The controller controls the electro-proportional valve to decrease the output air pressure, so that the contact force smoothly falls back to the target range.

[0031] Preferably, the adaptive model predictive control adopts a rolling time-domain optimization method. In each sampling period, the optimal control sequence is re-solved based on the latest state. Only the first control quantity of the sequence is output to the electro-proportional valve. In the next sampling period, the closed-loop process of state acquisition, model prediction, rolling optimization, and control quantity output is repeated. When the polishing tool first contacts the surface to be polished, the controller first compares the actual contact force with the target contact force, and then adjusts the input air pressure to avoid force overshoot and workpiece damage caused by contact impact.

[0032] Compared with the prior art, the present invention has the following beneficial effects:

[0033] 1. The present invention uses a fixed semi-cylindrical shell and a movable semi-cylindrical shell to form a slender cylindrical shape extending along the actuator axis and an internal cylindrical cavity covering both sides. The overall structure is slender and compact with a small radial dimension, which can enter narrow spaces such as holes, slots and deep cavities to perform internal surface polishing.

[0034] 2. This invention utilizes an axially connected direct transmission structure consisting of a pneumatic artificial muscle, an intermediate connector, a tension / compression sensor, a movable semi-cylindrical housing, and a polishing mechanism. This structure directly converts the axial tension of the pneumatic artificial muscle into the normal output force at the polishing tool end, eliminating the need for an internal linkage conversion mechanism, reducing the radial dimension, and simplifying the structure.

[0035] 3. The present invention uses two sets of spring guide components to provide axial constraint and pre-tightening support for the pneumatic artificial muscle. In the non-working state, it can balance the gravity of the moving side component and prevent the pneumatic artificial muscle from bending. In the working state, it can buffer force fluctuations and improve dynamic response.

[0036] 4. The present invention improves the overall rigidity of the actuator by using at least three sets of linear guide rail-slider assemblies, which can effectively limit torsion and sway during the polishing process and improve the stability of the narrow structure in the inner surface polishing scenario.

[0037] 5. This invention establishes a discrete prediction model for the actuator based on a three-element dynamic model that includes a contraction element, an elastic element, and a damping element. Combined with a parameter selection strategy based on air pressure range and tension or contraction state, and adaptive model predictive control, the model parameters are estimated and optimized online. This enables real-time adjustment of the input air pressure of the pneumatic artificial muscle, enhances the system's anti-interference ability, and improves the contact force control accuracy and polishing quality consistency. Attached Figure Description

[0038] Figure 1 is a schematic diagram of the overall structure of the actuator of the present invention;

[0039] Figure 2 is a cross-sectional view of the actuator of the present invention;

[0040] Figure 3 is a schematic diagram of the working principle of the actuator of the present invention in the deflation state, inflation state and working state;

[0041] Figure 4 is a partially enlarged structural schematic diagram of the spring guide assembly and force control mechanism of the present invention;

[0042] Figure 5 is a block diagram of the control system of the present invention;

[0043] Figure 6 is a flowchart of the working method of the present invention;

[0044] Figure 7 is a schematic diagram of the adaptive surface adjustment under curved surface conditions according to the present invention;

[0045] Reference numerals: 1. Flange mounting plate; 2. Fixed semi-cylindrical housing; 3. Movable semi-cylindrical housing; 4. Intermediate connector; 5. Tension / compression sensor; 6. Displacement sensor; 7. Pneumatic artificial muscle; 8. Guide shaft; 9. Linear bearing; 10. Bearing fixed end; 11. Compression spring; 12. Linear guide rail; 13. Slider; 14. Guide connector; 15. Drive motor; 16. Coupling; 17. Polishing tool; 18. Electro-proportional valve; 19. Air source interface; 20. Controller; 21. Industrial robot end flange; 22. Workpiece to be polished.

[0046] The present invention will be further described below with reference to the accompanying drawings and specific embodiments, but the scope of protection of the present invention is not limited to the following embodiments.

[0047] As shown in Figures 1 to 4, a force-controlled polishing actuator for the end effector of a robot based on pneumatic artificial muscles for polishing the inner surfaces of holes, grooves, and deep cavities includes a flange mounting plate 1, a fixed semi-cylindrical housing 2, a movable semi-cylindrical housing 3, an intermediate connector 4, a tension / compression sensor 5, a displacement sensor 6, a pneumatic artificial muscle 7, a guide shaft 8, a linear bearing 9, a bearing fixing end 10, a compression spring 11, a linear guide rail 12, a slider 13, a guide connector 14, a drive motor 15, a coupling 16, a polishing tool 17, an electro-proportional valve 18, an air source interface 19, and a controller 20.

[0048] The flange mounting plate 1 is disposed on the upper end of the fixed semi-cylindrical housing 2 and is used to connect with the end flange 21 of the industrial robot. The fixed semi-cylindrical housing 2 and the movable semi-cylindrical housing 3 are arranged opposite each other along the actuator axis. They cover each other in the circumferential direction to form a slender cylindrical shape and an internal cylindrical cavity covering both sides, thereby reducing the radial dimension of the actuator and improving accessibility in narrow spaces.

[0049] The movable semi-cylindrical housing 3 can move linearly along the actuator axis relative to the fixed semi-cylindrical housing 2 under the constraint of the guiding mechanism. The guiding mechanism is preferably a three-set linear guide rail slider assembly (including a linear guide rail 12 and a slider 13 sliding on the linear guide rail 12, the linear guide rail 12 can be fixed to the outer wall of the fixed semi-cylindrical housing 2, and the slider 13 can be fixed to the inner wall of the movable semi-cylindrical housing 3). The three sets of linear guide rail slider assemblies are evenly distributed between the fixed semi-cylindrical housing 2 and the movable semi-cylindrical housing 3 to limit lateral swing, torsion and offset, and improve the stiffness and stability of the actuator under long stroke conditions.

[0050] The force control mechanism is housed within the cylindrical cavity and includes a pneumatic artificial muscle 7, an intermediate connector 4, a tension / compression sensor 5, a displacement sensor 6, and two sets of symmetrically arranged spring guide assemblies. One end of the pneumatic artificial muscle 7 is fixed to the inner bottom of the fixed semi-cylindrical housing 2, and the other end is fixedly connected to one end of the intermediate connector 4. The other end of the intermediate connector 4 is connected to the force-receiving end of the tension / compression sensor 5, and the other end of the tension / compression sensor 5 is fixedly connected to the inner end face of the movable semi-cylindrical housing 3. The displacement sensor 6 is arranged parallel to the pneumatic artificial muscle 7, with its detection end facing the intermediate connector 4, and is used to detect the contraction displacement change of the pneumatic artificial muscle 7 during the polishing process. The two sets of spring guide assemblies are symmetrically arranged on both sides of the pneumatic artificial muscle 7 to ensure that it only moves along the axial direction and to suppress torsion.

[0051] The spring guide assembly is configured as two symmetrical groups, each group including a guide shaft 8, a linear bearing 9, a bearing fixing end 10, and a compression spring 11. The guide shaft 8 cooperates with the linear bearing 9 to ensure that the intermediate connecting piece 4 and its connected components move only along the actuator axial direction. Specifically, the bearing fixing end 10 is fixed to the inner wall of the fixed semi-cylindrical housing 2, and the linear bearing 9 is installed on the bearing fixing end 10; one end of the guide shaft 8 is fixed to the intermediate connecting piece 4, and the other end passes through the linear bearing 9 and can slide freely along the axial direction; the compression spring 11 is sleeved on the outside of the guide shaft 8, and its two ends abut against the end face of the linear bearing 9 and the side face of the intermediate connecting piece 4, respectively. The compression spring has a preset pre-compression amount in the initial assembly state of the actuator. When the actuator is not inflated, the pre-tightening force generated by the compression spring 11 is used to counteract the gravity of the moving side component and prevent the pneumatic artificial muscle 7 from bending under its own weight; when the actuator is working, the compression spring 11 can also buffer the output force changes caused by air pressure fluctuations and external disturbances.

[0052] The polishing mechanism is located at the lower end of the movable semi-cylindrical housing 3 and includes a drive motor 15, a coupling 16, and a polishing tool 17. The drive motor 15 is fixedly mounted on the movable semi-cylindrical housing 3, and its output shaft is connected to the polishing tool 17 through the coupling 16 to drive the polishing tool 17 to rotate and complete the polishing operation.

[0053] The air source interface 19 is connected to an external air source. An electric proportional valve 18 is located between the air source interface 19 and the pneumatic artificial muscle 7, and is used to adjust the input air pressure entering the pneumatic artificial muscle 7 according to the control signal output by the controller 20.

[0054] The controller 20 is electrically connected to the tension / compression sensor 5, the displacement sensor 6, and the electro-proportional valve 18, respectively, for real-time acquisition of contact force and displacement data, and outputs pressure regulation control signals to the electro-proportional valve 18 according to the control algorithm.

[0055] In this embodiment, the pneumatic artificial muscle 7, intermediate connector 4, tension / compression sensor 5, movable semi-cylindrical housing 3, and polishing mechanism are arranged sequentially along the actuator axis. Unlike existing structures that rely on internal linkage mechanisms to achieve force conversion, this invention does not provide an internal linkage conversion mechanism. Instead, it utilizes the aforementioned axial series direct transmission structure to directly transmit the axial tension generated when the pneumatic artificial muscle 7 contracts to the polishing tool 17, thereby forming the required normal contact force between the polishing tool 17 and the workpiece surface.

[0056] As shown in Figure 3, in the deflated state, the pneumatic artificial muscle 7 remains in an extended state and does not generate significant tension. At this time, the preload of the two sets of compression springs 11 is balanced with the total weight of the moving side components to maintain the initial position of the mechanism and prevent the pneumatic artificial muscle 7 from bending.

[0057] In the inflation state, the electric proportional valve 18 inputs compressed air into the pneumatic artificial muscle 7. Under the action of the input air pressure, the pneumatic artificial muscle 7 undergoes axial contraction and outputs tension. The intermediate connecting piece 4, the tension and pressure sensor 5, the movable semi-cylindrical housing 3, and the polishing mechanism move along the actuator axis under the action of this tension.

[0058] In operation, the polishing tool 17 contacts the surface of the workpiece to be polished. The workpiece surface provides a reaction support force to the polishing tool 17, and the actuator forms a normal contact force on the workpiece surface. Since the weight of the moving side component is a constant value, the spring force can be obtained from the spring compression and stiffness coefficient. Therefore, the contact force between the tool and the workpiece is mainly affected by the magnitude of the output tension of the pneumatic artificial muscle 7. By changing the air pressure input into the pneumatic artificial muscle 7, the axial tension of the pneumatic artificial muscle 7 can be adjusted, thereby adjusting the normal contact force between the polishing tool 17 and the workpiece surface to track the target value.

[0059] When the polishing tool 17 first contacts the workpiece surface, the controller 20 reads the actual contact force output by the tension / compression sensor 5. If the actual contact force is greater than the target contact force, the controller 20 reduces the output air pressure of the electro-proportional valve 18 to reduce the tension of the pneumatic artificial muscle 7; if the actual contact force is less than the target contact force, the controller 20 increases the output air pressure of the electro-proportional valve 18 to increase the tension of the pneumatic artificial muscle 7.

[0060] During the robot's movement along the workpiece surface, when a concave surface appears on the workpiece surface, the distance between the robot's end flange 21 and the surface to be polished increases, resulting in an increase in the contraction rate and a decrease in the tension of the pneumatic artificial muscle 7. The controller 20 increases the output air pressure of the electro-proportional valve 18 to maintain the set contact force. When a convex surface appears on the workpiece surface, the distance between the robot's end flange 21 and the surface to be polished decreases, resulting in a decrease in the contraction rate and an increase in the tension of the pneumatic artificial muscle 7. The controller 20 decreases the output air pressure of the electro-proportional valve 18 to maintain the set contact force.

[0061] As shown in Figures 5 to 7, a method for operating the above-mentioned actuator includes the following steps:

[0062] Step A1: Install the actuator onto the end effector of the industrial robot via flange mounting plate 1, power on both the actuator and the industrial robot, input initial air pressure into the pneumatic artificial muscle 7 through electro-proportional valve 18 to put the actuator into its initial state, and control the robot to move to the initial position so that the working end of the polishing tool 17 remains in a non-contact state with the workpiece surface.

[0063] Step A2: Input the target contact force, preset polishing trajectory and controller parameters into the controller 20, start the drive motor 15 to drive the polishing tool 17 to rotate, and control the electric proportional valve 18 to supply air to the pneumatic artificial muscle 7, so that the actuator enters the constant force control mode.

[0064] Step A3: The controller 20 acquires the contact force signal F(k) output by the tension and compression sensor 5 and the contraction displacement signal x(k) output by the displacement sensor 6 in real time, and calculates the displacement change, velocity or acceleration based on the contraction displacement signal to form the state quantity at the current sampling time; and constructs the model input quantity for online estimation and rolling prediction of model parameters based on the identified actuator dynamic model parameters.

[0065] In step A3 above, the controller 20 acquires the current output contact force, contraction displacement, and displacement change, velocity, or acceleration calculated from the contraction displacement through the tension / compression sensor 5 and the displacement sensor 6 (specifically, the controller 20 amplifies and converts the analog signal of the tension / compression sensor 5 to digital signal through a force transmitter, and acquires the digital displacement of the displacement sensor 6 through a displacement acquisition circuit; the contraction displacement signal is first processed by a first-order low-pass filter to remove noise, and then the velocity and acceleration are calculated through differential operation to improve the accuracy of the state quantities), and combines the identified actuator dynamic model parameters to form the model input quantities for online estimation and rolling prediction of model parameters.

[0066] Step A4: Establish a discrete prediction model based on a three-element dynamic model that includes a contraction element, an elastic element, and a damping element, and predict the state response and contact force response for several future steps in the rolling prediction time domain.

[0067] Preferably, the controller 20 selects the equivalent stiffness parameter in the three-element dynamic model according to the preset pressure range to which the current input air pressure belongs, and determines whether the actuator is in a stretched state or a contracted state according to the direction of change of the displacement in the current sampling period relative to the displacement in the previous sampling period, so as to determine the equivalent damping parameter, thereby improving the model's adaptability to the nonlinearity and hysteresis characteristics of the pneumatic artificial muscle.

[0068] In one embodiment, different equivalent stiffness parameter relationships can be selected based on whether the current input air pressure is in the range of 1 bar to 3 bar or 3 bar to 6 bar; and different equivalent damping parameter relationships can be selected based on the direction of change of the displacement in the current sampling period relative to the displacement in the previous sampling period.

[0069] Step A5: The controller 20 estimates and updates the parameters of the discrete prediction model online based on the deviation between the predicted output and the actual output, and in conjunction with the current state observation (a recursive least squares algorithm can be used); based on the updated discrete prediction model, the target contact force and the constraints (such as pressure output limit, pneumatic artificial muscle stroke limit, contact force change rate limit, etc.), it solves the optimal control sequence and takes the first term of the optimal control sequence as the control pressure p(k) at the current sampling time.

[0070] Step A6: Controller 20 outputs control air pressure p(k) to electro-proportional valve 18 to adjust the input air pressure entering the pneumatic artificial muscle 7, so that the polishing tool 17 maintains a set normal contact force and completes polishing when moving along the inner surface of holes, grooves and deep cavities under the drive of the robot. Preferably, the adaptive model predictive control uses a rolling optimization method to obtain the future control sequence, and only uses the first control quantity of the current sampling period as the actual input of electro-proportional valve 18.

[0071] This invention's actuator can be used not only for polishing the outer surface of workpieces, but also for polishing the inner surface of space-constrained areas such as holes, grooves, and deep cavities. Through its small radial dimension, long axial structure, and axial series direct-drive force control scheme, it can access internal areas of workpieces that are difficult to reach with traditional end effectors.

[0072] In practical applications, the industrial robot drives the actuator to enter the workpiece hole, groove or deep cavity according to the preset trajectory. The polishing tool 17 contacts the inner surface in the rotating state. The controller 20 adjusts the output air pressure of the electric proportional valve 18 through the real-time feedback of the tension and pressure sensor 5 and the displacement sensor 6, so that the polishing tool 17 always maintains a constant normal contact force on the inner surface of the workpiece, thereby improving the polishing uniformity and surface quality.

[0073] It should be noted that the above embodiments are only preferred embodiments of the present invention. Equivalent substitutions or conventional changes made by those skilled in the art to the structural form, number of components, installation position, control process or parameter settings without departing from the essential spirit of the present invention shall all fall within the protection scope of the present invention.

Claims

1. A robotic force-controlled actuator for deep cavity internal polishing, characterized in that: The device includes a flange mounting plate (1), a fixed semi-cylindrical housing (2), a movable semi-cylindrical housing (3), a force control mechanism, a polishing mechanism, an air source interface (19), an electro-proportional valve (18), and a controller (20). The flange mounting plate (1) is fixedly installed at one end of the fixed semi-cylindrical housing (2) near the industrial robot and is used for detachable connection with the end flange (21) of the industrial robot. The fixed semi-cylindrical housing (2) and the movable semi-cylindrical housing (3) are sleeved relative to each other along the actuator axis. They are mutually wrapped and fitted in the circumferential direction to form a slender cylindrical shape extending along the actuator axis, and the interior is enclosed to form a cylindrical cavity with two sides. The force control mechanism is housed in the cylindrical cavity. Inside the cavity, there are pneumatic artificial muscles (7), intermediate connectors (4), tension and compression sensors (5), displacement sensors (6), and two sets of symmetrically arranged spring guide assemblies; one end of the pneumatic artificial muscle (7) is fixed to the inner bottom of the fixed semi-cylindrical housing (2), and the other end is fixedly connected to one end of the intermediate connector (4); the other end of the intermediate connector (4) is connected to the force-bearing end of the tension and compression sensor (5), and the other end of the tension and compression sensor (5) is fixedly connected to the inner end face of the movable semi-cylindrical housing (3); the displacement sensor (6) is arranged parallel to the pneumatic artificial muscle (7), the fixed end of the displacement sensor (6) is installed inside the fixed semi-cylindrical housing (2), and the detection end faces the intermediate connector (4). The system is configured to collect the axial contraction displacement of the pneumatic artificial muscle (7) in real time; two sets of spring guide assemblies are symmetrically distributed on both radial sides of the pneumatic artificial muscle (7), and their ends are respectively connected to the fixed semi-cylindrical housing (2) and the intermediate connecting piece (4); the polishing mechanism is fixedly installed on the outer end of the movable semi-cylindrical housing (3) away from the tension and compression sensor (5), and moves synchronously with the movable semi-cylindrical housing (3) along the actuator axis; the air source interface (19) is set on the side wall of the fixed semi-cylindrical housing (2) for connecting to an external compressed air source; the electric proportional valve (18) is connected in series in the air path between the air source interface (19) and the pneumatic artificial muscle (7) for precisely adjusting the input pneumatic artificial muscle. The air pressure value of the muscle (7); the controller (20) is electrically connected to the tension and pressure sensor (5), the displacement sensor (6) and the electric proportional valve (18) respectively; wherein, the pneumatic artificial muscle (7), the intermediate connector (4), the tension and pressure sensor (5), the movable semi-cylindrical housing (3) and the polishing mechanism are connected in series along the actuator axis to form an axial series direct transmission force transmission chain without steering transmission links; the actuator does not have a linkage conversion mechanism for changing the direction of force transmission, and the axial tension generated by the inflation and contraction of the pneumatic artificial muscle (7) is directly transmitted to the polishing mechanism through the axial series direct transmission force transmission chain, so that the working end of the polishing mechanism forms a normal output contact force on the polishing surface of the workpiece (22) to be polished.

2. The robotic force-controlled actuator for deep cavity internal polishing according to claim 1, characterized in that, Each set of spring guide assemblies includes a guide shaft (8), a linear bearing (9), a bearing fixing end (10), and a compression spring (11); the bearing fixing end (10) is fixed to the inner wall of the fixed semi-cylindrical housing (2), and the linear bearing (9) is installed on the bearing fixing end (10); one end of the guide shaft (8) is fixed to the intermediate connecting member (4), and the other end passes through the linear bearing (9) and can slide freely along the axial direction; the compression spring (11) is sleeved on the outside of the guide shaft (8), and both ends abut against the end face of the linear bearing (9) and the side face of the intermediate connecting member (4), respectively; the compression spring (11) has a preset pre-compression amount in the initial assembly state of the actuator, so as to balance the gravity of the moving side component in the non-working state of the actuator, and buffer the output force fluctuation in the working state.

3. The robotic force-controlled actuator for deep cavity internal polishing according to claim 1, characterized in that, The actuator is also equipped with a guide mechanism, which includes at least three sets of linear guide slider assemblies. Each set of linear guide slider assemblies is evenly distributed along the circumference of the actuator, and the linear guide (12) of the linear guide slider assembly is fixed axially to the outer wall of the fixed semi-cylindrical housing (2), and the slider (13) of the linear guide slider assembly is correspondingly fixed to the inner wall of the movable semi-cylindrical housing (3), which is used to improve the overall rigidity of the actuator and limit the torsion and sway during the polishing process.

4. The robotic force-controlled actuator for deep cavity internal polishing according to claim 1, characterized in that, The polishing mechanism includes a drive motor (15), a coupling (16), and a polishing tool (17). The body of the drive motor (15) is fixedly installed on the movable semi-cylindrical housing (3). The output shaft of the drive motor (15) is coaxially connected to the clamping end of the polishing tool (17) through the coupling (16) to drive the polishing tool (17) to rotate around its own axis. The rotation axis of the polishing tool (17) coincides with the axial center line of the actuator to ensure that the polishing contact force is output uniformly along the normal direction.

5. A method for controlling a robot force actuator for deep cavity internal polishing, characterized in that, The control method using the force-controlled actuator according to any one of claims 1 to 4 includes the following steps: Step A1: Fix the force control actuator to the end of the industrial robot via the flange mounting plate (1), and power on both the force control actuator and the industrial robot; input the preset initial air pressure to the pneumatic artificial muscle (7) through the electro-proportional valve (18) to keep the actuator in its initial state; control the industrial robot to move the actuator to the polishing initial point, so that the working end of the polishing tool (17) and the surface of the workpiece (22) to be polished maintain a preset safe distance without contact; Step A2: Input the target contact force value, preset polishing trajectory parameters and controller initial parameters to the controller (20); start the drive motor (15) of the polishing mechanism to drive the polishing tool (17) to rotate, and at the same time control the electric proportional valve (18) to supply air to the pneumatic artificial muscle (7) so that the actuator enters the constant force closed loop control mode; Step A3: The controller (20) collects the contact force signal F(k) output by the tension and compression sensor (5) and the contraction displacement signal x(k) output by the displacement sensor (6) in real time with a fixed sampling period; the displacement change, motion velocity and motion acceleration at the current sampling time are calculated based on the contraction displacement signal x(k), which together constitute the current system state variables; combined with the initial parameters of the actuator dynamics model obtained offline, a model input vector for online estimation of model parameters and rolling prediction is constructed. Step A4: Based on the three-element dynamic model including the contraction element, elastic element and damping element, establish the discrete prediction model of the actuator; within the set rolling prediction time domain, based on the current system state variables and the model input vector, predict the state response and contact force output response of multiple future sampling steps; Step A5: Calculate the deviation between the predicted contact force response and the actual collected contact force signal. Combined with the current system state observations, use the recursive least squares algorithm to estimate and update the parameters of the discrete prediction model online and in real time. Based on the updated discrete prediction model, with the target contact force as the tracking target, and combined with the constraints of air pressure output limit, pneumatic artificial muscle stroke limit, and contact force change rate limit, solve the optimal control air pressure sequence in a rolling manner, and take the first term of the optimal control sequence as the control air pressure p(k) at the current sampling time. Step A6: The controller (20) converts the control air pressure p (k) into a corresponding control signal and outputs it to the electric proportional valve (18) to adjust the input air pressure of the pneumatic artificial muscle (7) in real time. During the process of the industrial robot driving the actuator to move along the inner surface to be polished, the normal contact force between the polishing tool (17) and the inner surface of the workpiece is kept stable at the target value through closed-loop control, so as to complete the constant force polishing operation of the inner surface of the deep cavity.

6. The robot force-controlled actuator control method for deep cavity internal polishing according to claim 5, characterized in that, In step A3, the controller (20) amplifies and converts the analog signal of the tension and compression sensor (5) to digital through the force transmitter, and obtains the digital displacement of the displacement sensor (6) through the displacement acquisition circuit; the contraction displacement signal x(k) is first processed by first-order low-pass filtering to remove noise, and then the velocity and acceleration are calculated by differential operation to improve the accuracy of the state quantity.

7. The robot force-controlled actuator control method for deep cavity internal polishing according to claim 5, characterized in that, In step A4, the controller (20) first determines the preset pressure range to which the input air pressure of the current pneumatic artificial muscle (7) belongs, and matches the corresponding equivalent stiffness parameter according to the pressure range; then compares the contraction displacement value of the current sampling period with that of the previous sampling period to determine whether the actuator is currently in the contraction stroke or the extension stroke, and matches the corresponding equivalent damping parameter according to the stroke direction; and substitutes the matched equivalent stiffness and equivalent damping parameters into the three-element dynamic model to update the parameter matrix of the discrete prediction model in order to adapt to the nonlinear and hysteresis characteristics of the pneumatic artificial muscle.

8. The robot force-controlled actuator control method for deep cavity internal polishing according to claim 5, characterized in that, In step A6, the surface adaptive adjustment logic of the actuator when it moves along the inner surface of the curved surface with the robot is as follows: when the surface to be polished is concave and the distance between the robot end flange and the surface to be polished increases, the contraction rate of the pneumatic artificial muscle (7) increases, the output tension decreases, and the contact force is lower than the target value. The controller (20) controls the electric proportional valve (18) to increase the output air pressure, so that the contact force quickly rises back to the target range. When the surface to be polished is convex and the distance between the robot end flange and the surface to be polished decreases, the contraction rate of the pneumatic artificial muscle (7) decreases, the output tension increases, and the contact force is higher than the target value. The controller (20) controls the electric proportional valve (18) to decrease the output air pressure, so that the contact force smoothly falls back to the target range.

9. The robot force-controlled actuator control method for deep cavity internal polishing according to claim 5, characterized in that, The adaptive model predictive control adopts a rolling time-domain optimization method. In each sampling period, the optimal control sequence is re-solved based on the latest state. Only the first control quantity of the sequence is output to the electric proportional valve (18). In the next sampling period, the closed-loop process of state acquisition, model prediction, rolling optimization, and control quantity output is repeated. When the polishing tool first contacts the surface to be polished, the controller first compares the actual contact force with the target contact force, and then adjusts the input air pressure to avoid force overshoot and workpiece damage caused by contact impact.