Parallel platform anti-interference decoupling control system based on multi-sensor fusion

Abstract

The invention discloses a parallel platform anti-interference decoupling control system based on multi-sensor fusion, and relates to the technical field of pose control. According to the system, a comprehensive state sensing network is constructed by integrating three types of sensors, namely a first sensing unit for measuring the motion amount of a driving joint, a second sensing unit for detecting the output force of the driving joint and a third sensing unit for measuring the inertial motion of a movable platform. The controller operates three function decoupling control loops in parallel based on multi-source sensing information: a main tracking loop realizes macroscopic trajectory tracking based on a kinematic model and generates a reference output instruction; the inertia feedback loop generates feedback force to compensate dominant inertia dynamics by using the inertia information of the operation space; and the force interference feedback loop generates compensation force according to the deviation between the joint output force and the reference output instruction, and suppresses unmodeled dynamic and external disturbance. According to the invention, through multi-loop cooperation, high-performance anti-interference control independent of an accurate global dynamic model is realized.

Description

Technical Field

[0001] This invention relates to the field of posture control technology, specifically to a parallel platform anti-interference decoupling control system based on multi-sensor fusion. Background Technology

[0002] In the field of high-precision pose stabilization control of parallel platforms, especially in the continuously disturbed operating environment of basic platforms such as ships and vehicles, effectively suppressing the dual disturbances caused by basic vibration and end-load changes has always been a core technical challenge. Traditional solutions mainly follow two technical routes. First, they rely heavily on accurate global dynamic models, using feedback compensators or state observers (such as Kalman filters) to estimate and cancel disturbances. However, the strong coupling and nonlinear dynamic characteristics of parallel mechanisms themselves make accurate modeling extremely difficult. Model parameters are not only volatile but also difficult to obtain, leading to model-based control strategies often introducing new errors in practical applications due to model mismatch. In addition, the computational complexity results in high engineering implementation costs. Second, to circumvent the modeling difficulties, some solutions adopt pure kinematic control combined with traditional PID feedback. While these methods simplify system design to some extent, their response inherently lags. For high-frequency, high-dynamic disturbances such as wave impacts, correction often only initiates after the disturbance has accumulated into a significant pose error, leading to tracking lag and system overshoot. This results in the end-effector being subjected to multiple impacts, making it difficult to meet the requirements of high-precision and stable operation. Although there have been attempts to introduce force sensors in existing technologies, their potential has not been fully explored. Traditional methods have failed to construct an efficient closed loop based on direct measurement of disturbance forces, and cannot achieve fast and accurate disturbance suppression while bypassing complex models. Therefore, existing technologies generally suffer from inherent contradictions between model dependence and real-time performance, and between control accuracy and engineering simplicity. There is an urgent need for an innovative control strategy that can break free from the constraints of precise dynamic models and provide fast and direct compensation for multi-source disturbances. Summary of the Invention

[0003] To break free from the constraints of precise dynamic models and to provide rapid and direct compensation for multi-source disturbances, this invention proposes a parallel platform anti-interference decoupling control system based on multi-sensor fusion. The system includes a parallel mechanism, a multi-source sensing system, and a controller. The parallel mechanism comprises a fixed platform, a moving platform, and multiple drive chains connecting the two. The multi-source sensing system includes: a first sensing unit disposed on each drive branch for acquiring the actual motion of the drive joint in the joint space; a second sensing unit disposed on each drive branch for detecting the actual output force of the drive joint in the joint space; and a third sensing unit disposed on the moving platform for measuring the inertial motion information in the operating space of the moving platform. By fusing information from multiple sensor systems, the controller is configured to execute three functionally decoupled control loops in parallel. It generates control commands to drive the parallel mechanism by fusing information from the multiple sensor systems. The control loops include: The main tracking loop obtains the expected motion of the driving joint based on the desired trajectory through inverse kinematics solution, and compares it with the actual motion to generate joint space error compensation. At the same time, it obtains the operation space error based on inertial motion information and maps it to the joint space compensation amount, and then superimposes it to achieve the reference output command for macroscopic motion tracking. The force disturbance feedback loop generates a compensating force to suppress unmodeled dynamics and external disturbances based on the deviation between the actual output force and the reference output force command. An inertial feedback loop, based on inertial motion information in the operating space, generates a feedback force to compensate for unmodeled inertial dynamics by mapping the inertial model to the force Jacobian and comparing it with the actual output force.

[0004] Furthermore, the first sensing unit is an encoder built into the drive motor of the drive joint, and the measured actual motion is the linear displacement and / or motion speed of the push rod at the corresponding joint.

[0005] Furthermore, the second sensing unit is a single-dimensional force sensor installed in series in the force flow path of the drive branch, and the central detection axis of the single-dimensional force sensor and the thrust axis of the drive branch are configured to remain collinear, and the detected actual output force is the output force along the joint axis.

[0006] Furthermore, the unidimensional force sensor is connected to the drive chain via a preload bolt.

[0007] Furthermore, the force interference feedback loop includes a low-pass filter, which filters the actual output force measured by the second sensing unit to suppress high-frequency bending moment interference generated by the swing of the drive branch.

[0008] Furthermore, the third sensing unit is an inertial measurement unit rigidly fixed to the moving platform, and the measurement axis of the inertial measurement unit is aligned with the working coordinate system of the moving platform. The measured spatial inertial motion information is the linear acceleration and angular acceleration of the moving platform.

[0009] Furthermore, the inertial model upon which the inertial feedback loop relies is a simplified model based on the mass and rotational inertia of the moving platform.

[0010] Furthermore, the feedback force in the inertial feedback loop is generated through the following steps: Based on the inertial motion information measured by the third sensing unit, and combined with the inertial matrix formed by the equivalent mass and moment of inertia of the moving platform, the inertial force in the operating space is calculated. By transposing the force Jacobian matrix of the parallel mechanism, the inertial force in the operating space is mapped to the joint space, and the difference between the inertial force and the actual output force is used to generate a feedback force to compensate for the unmodeled inertial dynamics.

[0011] Furthermore, the main tracking loop includes a proportional-derivative controller, which is configured to generate a reference output command based on the deviation between the expected and actual joint motion.

[0012] Compared with the prior art, the present invention has at least the following beneficial effects: This invention proposes a parallel platform anti-interference decoupling control system based on multi-sensor fusion. It constructs a multi-source sensing network composed of three types of sensors: joint space motion, joint space output force, and operational space inertial motion information. Based on this network, three functionally decoupled control loops are deployed in parallel within the controller, ultimately forming a high-performance control paradigm fundamentally different from traditional model-dependent paths. This system achieves "decoupling" and "directness" in anti-interference control. Specifically, the system ensures stable tracking of the macroscopic trajectory through the main tracking loop, establishing a control benchmark; using an inertial feedback loop, with the aid of a simplified inertial model and force Jacobian mapping, it provides feedback compensation for the system-dominant inertial dynamic effects, achieving initial decoupling at the dynamic level; crucially, the force interference feedback loop constructs a local force closed loop by directly comparing the measured values ​​of joint space forces with the desired commands, enabling real-time sensing and direct cancellation of all residual interferences, including unmodeled dynamics, branch couplings, and external impacts. The coordinated operation of these three loops allows the system to effectively cope with multi-source interferences such as foundation vibration and load changes without relying on a precise and complex global nonlinear dynamic model. Attached Figure Description

[0013] Figure 1 This is an anti-interference decoupling control system based on a parallel platform using multi-sensor fusion; Figure 2 This is a schematic diagram of the 3-SPR configuration; Figure 3 An enlarged view of the P joint in the SPR configuration; Figure 4 For the control method block diagram; Explanation of reference numerals in the attached drawings: 100-fixed platform, 200-moving platform, 300-SPR drive chain, 301-ball joint, 302-cylinder connecting rod, 303-counterweight, 304-electric cylinder actuator, 305-rotary joint, 400-inertial measurement unit (IMU), 500-single-dimensional force sensor. Detailed Implementation

[0014] The following are specific embodiments of the present invention, which are described in conjunction with the accompanying drawings. However, the present invention is not limited to these embodiments.

[0015] The core of this invention lies in providing a control paradigm that does not rely on a precise global dynamic model. Through the functional coordination and information fusion of multiple source sensors, multiple parallel and decoupled control loops are constructed to address trajectory tracking, inertial dynamics compensation, and unknown disturbance suppression, respectively. This enables high-precision and robust attitude stability control of the moving platform under complex operating conditions with underlying vibration and external load variations. Figure 1 As shown, this invention proposes a parallel platform anti-interference decoupling control system based on multi-sensor fusion, including a parallel mechanism, a multi-source sensing system, and a controller. The parallel mechanism includes a fixed platform, a moving platform, and multiple drive chains connecting the two. The multi-source sensing system includes: a first sensing unit disposed on each drive branch for acquiring the actual motion of the drive joint in the joint space; a second sensing unit disposed on each drive branch for detecting the actual output force of the drive joint in the joint space; and a third sensing unit disposed on the moving platform for measuring the inertial motion information in the operating space of the moving platform. By fusing information from multiple sensor systems, the controller is configured to execute three functionally decoupled control loops in parallel. It generates control commands to drive the parallel mechanism by fusing information from the multiple sensor systems. The control loops include: The main tracking loop obtains the expected motion of the driving joint based on the desired trajectory through inverse kinematics solution, and compares it with the actual motion to generate joint space error compensation. At the same time, it obtains the operation space error based on inertial motion information and maps it to the joint space compensation amount, and then superimposes it to achieve the reference output command for macroscopic motion tracking. The force disturbance feedback loop generates a compensating force to suppress unmodeled dynamics and external disturbances based on the deviation between the actual output force and the reference output force command. An inertial feedback loop, based on inertial motion information in the operating space, generates a feedback force to compensate for unmodeled inertial dynamics by mapping the inertial model to the force Jacobian and comparing it with the actual output force.

[0016] First, the system hardware configuration of this invention will be described in detail. For example... Figure 2As shown, in a preferred embodiment, the parallel mechanism adopts a 3-SPR configuration, which provides an ideal mechanical structural foundation for achieving stable control with high stiffness and low inertia. This mechanism includes a fixed platform (100) serving as the mounting base, a moving platform (200) for carrying the working tools, and three identical SPR drive chains (300) connecting the fixed platform (100) and the moving platform (200). In practical applications, the fixed platform (100) is typically bolted to a vibrating base, such as a ship's deck or vehicle chassis; while the moving platform (200), as the ultimate target for anti-interference control, requires maintaining high-precision positional stability.

[0017] Furthermore, each SPR drive chain (300) comprises a series of precision components that together form the basis for the system's execution and sensing. Specifically, the chain begins with a ball joint (301) mounted on a fixed platform (100), which provides a critical spatial rotation center. Connected below the ball joint (301) is a cylinder connecting rod (302), the top of which is connected to the lower end of the ball joint rod of the ball joint (301) via a connecting flange. In this connection, a key single-dimensional force sensor (500), as the second sensing unit of the invention, is installed in series between the ball joint (301) and the cylinder connecting rod (302). This installation position is located in the force flow path of the drive chain, ensuring accurate detection of the actual output force of the drive joint. The installation accuracy here is crucial; it is essential to ensure a high degree of coaxiality among the central detection axis of the single-dimensional force sensor (500), the thrust axis of the cylinder connecting rod (302), and the theoretical force transmission axis of the ball joint (301). To achieve this, the flange connection at both ends of the sensor must be secured with pre-tightened bolts to maximize connection rigidity and prevent the introduction of unnecessary flexible vibrations that could contaminate the force signal quality. Notably, a counterweight (303) can be placed near the ball joint (301) to balance the position of the electric cylinder's center of gravity; this design detail helps improve the overall dynamic characteristics of the system.

[0018] Specifically, the core of the single-dimensional force sensor (500) in the single SPR-type drive branch (300) lies in its direct integration with the P joint (i.e., the prism joint, the drive end of P). In the enlarged view section ( Figure 3As can be seen, the unidirectional force sensor (500) is rigidly connected in series in the drive chain via mounting flanges at both ends: its lower end is connected to the ball joint (301), and its upper end is directly fixed to the bottom of the housing of the electric cylinder actuator (304). Ignoring the external housing, the cylindrical push rod of the electric cylinder extends from the center of the motor, and its axis is strictly coaxial with the central detection axis of the unidirectional force sensor (500). The core purpose of this design is to ensure that the thrust generated by the motor that moves the drive joint P is transmitted synchronously to the sensitive element of the unidirectional force sensor (500) before being transmitted to the R joint (revolute joint). Therefore, the axial force measured by the sensor is the true, original output force applied to the chain by the drive joint P, without any loss due to intermediate links. This measurement directly reflects the actual execution of the drive command, providing the most direct and reliable source of physical information for subsequent force feedback control. To ensure measurement accuracy, the coaxiality error between the force sensor (500) and the push rod must be controlled within a very small range during installation. This will minimize measurement errors caused by off-center loading and ensure that the sensor only senses pure axial force.

[0019] Based on this, the core drive component of the SPR drive chain (300), the electric cylinder actuator (304), is nested within the cylinder connecting rod (302). The electric cylinder actuator (304) consists of multiple components working in concert: a permanent magnet synchronous linear motor as a power source provides the initial driving force; a push rod driven by the motor through a ball screw achieves precise displacement; and a high-precision encoder built into the electric cylinder serves as the first sensing unit of this invention, used to measure the linear displacement and velocity of the push rod relative to the cylinder in real time. The end of the push rod is connected to the moving platform (200) through a revolute joint (305), which is carefully designed to release only the rotational degree of freedom of the winding contact, thus forming, together with the ball joint joint (301) of the fixed platform (100), the complete motion constraint chain required by the SPR drive chain (300).

[0020] After describing the SPR drive chain (300), the operational space sensing configuration of the system needs to be emphasized. An inertial measurement unit (IMU) (400), as the third sensing unit of this invention, is rigidly fixed to the center of the moving platform (200) via a dedicated mounting plate. During installation, it must be ensured that there is no relative movement between the housing of the IMU (400) and the moving platform (200), and that its measurement plane is parallel to the working plane of the moving platform (200). At the same time, each measurement axis of the IMU (400) must be precisely aligned with the preset inertial reference system or working coordinate system of the moving platform (200). These strict installation requirements ensure that the measured inertial motion information can truly and accurately reflect the actual motion state of the moving platform (200) in the operational space.

[0021] After a complete description of the system's hardware configuration, the control method of this invention will be elaborated upon. The control method of this invention is the intelligent core of the system. Its design strictly adheres to the sensing capabilities provided by the aforementioned hardware configuration. The entire control architecture, without relying on complex nonlinear dynamic models, achieves high-precision anti-interference decoupled control through multi-loop fusion and compensation of position, force, and inertial information. In specific implementation, the controller is typically implemented by an industrial computer or embedded system with high-speed computing capabilities. It is configured to execute three functionally decoupled but informationally coordinated control loops in parallel: a main tracking loop, a force disturbance feedback loop, and an inertial feedback loop. The outputs of these three loops are ultimately fused at the current command level to jointly drive the actuator.

[0022] Before delving into the individual control loops, it is essential to first establish the system's kinematic foundation. To this end, the system requires the construction of an inverse kinematic model of the 3-SPR parallel mechanism. This model is built upon the platform's geometry, including key parameters such as the hinge point distribution radius of the moving and stationary platforms and the hinge point position angles. This model accurately describes the mathematical mapping relationship between the operating space and the joint space. For the 3-SPR mechanism, its moving platform can typically be simplified to have vertical displacement. , around Axis rotation angle and rotation angle around the Y-axis Three degrees of freedom. Therefore, as... Figure 4 The diagram shown is a block diagram of the control method of the system proposed in this invention. The operation space pose can be represented as a vector. The core task of the inverse kinematics model is to determine this desired pose. The expected elongation of the three electric cylinder push rods is mapped to form the expected displacement vector in the joint space. Its mathematical model can be expressed as: And thus, we can deduce .in, For the reason and The attitude rotation matrix of the moving platform relative to the fixed platform. Let be the position vector of the center of mass of the moving platform relative to the center of mass of the stationary platform. and These are the distribution position vectors of the hinge points on the moving platform and the fixed platform, respectively.

[0023] The main tracking loop is the foundation for the system to achieve macroscopic trajectory tracking; it forms the framework of the entire control system. This loop is initiated based on the desired trajectory of the moving platform. Using the inverse kinematics model described above, the controller calculates the corresponding desired joint displacements in real time. At the same time, the controller reads the actual push rod displacement fed back by the built-in encoders of each electric cylinder. This establishes a precise perception of the system's actual state. Subsequently, the system calculates the position tracking error for each drive chain. The error is input to a proportional-derivative controller, whose control law is: .in, This refers to the reference output command generated by the circuit for achieving macroscopic motion tracking, measured in Newtons. and These are the proportional gain matrix and the differential gain matrix, typically 3×3 positive definite diagonal matrices. Their main diagonal elements determine the system's response stiffness to position errors and the damping effect it provides, respectively. An IMU (Inertial Measurement Unit) mounted on the moving platform measures the platform's acceleration and angular velocity in real time. Through integration and filtering fusion (e.g., fusing encoder and IMU data using complementary filtering or Kalman filtering algorithms), the actual pose of the moving platform in inertial space is calculated. According to the desired pose Calculate the operating space error Generate operational space pose error The pose error in the operating space is calculated by using the inverse of the velocity Jacobian matrix J of the 3-SPR parallel mechanism. Mapped to joint space, a displacement compensation amount is generated. ; Displacement compensation amount Compared with the expected displacement calculated by inverse kinematics They are superimposed and used together as a new desired displacement command. This conversion is achieved through the force constant of the electric cylinder motor. (Unit: N / A) Converted to a basic current command ,Right now (in This is the desired current value input to the current loop of the electric cylinder servo driver.

[0024] However, although the main tracking loop can achieve basic trajectory tracking, it is essentially a hysteresis compensation based on position error, and cannot sense or compensate for branch inertia, joint friction, and complex dynamic coupling effects. Especially for high-frequency disturbances, the position loop needs to wait for error accumulation before responding, resulting in an inherent delay. To overcome this limitation, this invention creatively introduces an inertial feedback loop and a force disturbance feedback loop, together forming a multi-layered, forward-looking anti-interference system.

[0025] The force interference feedback loop, a key component for achieving high-frequency anti-interference in this system, constructs a local force closed loop independent of the position loop, specifically designed to suppress remaining unmodeled dynamics and external disturbances. In each control cycle, a single-dimensional force sensor connected in series in the electric cylinder's thrust path directly measures the actual force exerted by the push rod on the base. At the signal processing level, the raw signal first needs to be precisely calibrated and filtered. After the system is installed, a rigorous sensor calibration procedure must be performed: the output voltage is collected as the zero-point value when the electric cylinder is unloaded. Subsequently, in current loop mode, a series of known currents were input and the corresponding output voltages were recorded. This allows us to calculate the calibration coefficients of the sensor. (Unit: N / V), and through The original force signal is obtained. Due to actual factors such as branch motion, hinge clearance, and structural deformation, the measured value... This will inevitably include additional bending moment disturbances caused by oscillation, which typically manifest as high-frequency noise. Therefore, a low-pass filter is needed in the software to filter these disturbances. Low-pass filtering is applied to extract joint spatial force signals. To further eliminate the influence of systematic bending moment under specific working conditions, the system can also allow the electric cylinder to reciprocate at low speed throughout its entire range under safe conditions without external load, recording the characteristic curves of force sensor readings as a function of displacement and velocity. And in real-time control, this fitted value is subtracted from the filtered signal, i.e. This allows us to obtain a purer force signal that truly reflects external interference.

[0026] This circuit processes the measured joint output force. The reference output command generated by the main tracking loop Real-time comparison is performed, and the difference is... This is considered a generalized disturbance force comprised of unmodeled dynamics (excluding inertia), branch coupling, and external disturbances. The force is input to a dedicated force controller (which can be proportional or integral) to generate the corresponding compensating force, which is ultimately converted into a compensating current. .

[0027] The inertial feedback loop aims to identify and compensate for unmodeled inertial dynamics arising from model simplification in the system, achieving precise decoupling of dynamics. The loop's design is based on the understanding that, due to the simplified inertial model, there is a discrepancy between the calculated theoretical inertial force and the system's actual inertial force. Therefore, a mechanism is needed to observe and compensate for this discrepancy. This loop heavily relies on real-time data provided by the IMU on the moving platform. The IMU continuously measures the linear and angular accelerations of the moving platform in the operating space, denoted as... The controller is based on a simplified inertial model that relies only on the easily obtainable or estimable mass m of the moving platform and the rotational distance. Moment of inertia of the shaft Constructing the inertia matrix Using this matrix and measured acceleration, the operational space inertial force required to drive the platform to generate the current acceleration under an ideal model can be accurately calculated. .

[0028] Subsequently, through the Liyakubi matrix of the parallel mechanism transpose Based on the principle of virtual work, the inertial force in this operating space is mapped to the joint space to generate an ideal inertial force expectation value for the joint space. .

[0029] However, due to model simplification and the presence of unmodeled dynamics, the actual inertial force deviates from the ideal value. To compensate for this deviation, the loop introduces direct measurements from force sensors for comparison. The controller reads the filtered actual output force of each electric cylinder. This is compared with the calculated expected value of the theoretical inertial force. The comparison is then performed. This comparison step is the core of the inertial feedback loop, and the difference is defined as: This difference This intuitively reflects the unmodeled inertial dynamics effect caused by model mismatch. The difference is then... The input is fed into a compensation controller (such as a proportional or integral controller) to generate a feedback force for accurately compensating for the unmodeled inertial dynamics, which is then ultimately converted into a compensation current. This mechanism enables the active observation and compensation of model uncertainties in the dominant inertial dynamics.

[0030] Finally, the controller intelligently merges the current commands generated by the three loops to obtain the total current control command. This final command is sent in real time to the servo drives of each electric cylinder for execution, thereby precisely driving the parallel mechanism. Through this multi-loop cooperative mechanism, the system not only ensures accurate tracking of the macroscopic trajectory, but also compensates for the dominant dynamics through inertial feedback. Finally, it uses a high-frequency force closed loop to cancel out all residual disturbances in real time and directly, forming a complete and efficient control closed loop.

[0031] In contrast, under traditional position control without introducing feedback decoupling, the closed-loop system of the 3-SPR parallel platform can be simplified into a multiple-input multiple-output (MIMO) system. The relationship between its output (platform pose) and input (electric cylinder current) is dominated by the internal coupling dynamics of the system.

[0032] While its independent PD controllers for each branch can effectively compensate for position errors caused by internal coupling interference on a stable platform, this coupling effect is amplified in a marine environment, introducing more interference from unspecified sources. These include ship vibration and movement, and sudden wave changes, which complicate the relationship between input and output, as detailed below: set up Let be the transfer function matrix of the controlled object (3-SPR mechanism). It is a 3×3 coupling matrix with non-zero diagonal elements, representing interference coupling between branches. The system relationship is as follows: , in, It is the Laplace transform of the motion of the electric cylinder push rod. It is the Laplace transform of the input current command.

[0033] , Coupling matrix off-diagonal elements , This represents the influence of the input current of the j-th electric cylinder on the motion of the push rod of the i-th electric cylinder under interference conditions, represented by the unit loop controller matrix. off-diagonal elements This represents the composite error between the various drivers, amplified continuously under interference conditions, as well as the composite coefficient of additional interference. This coupling is determined by the complex mechanical structure of the parallel mechanism and the multi-source interference of the complex environment.

[0034] In this situation, any external disturbance experienced by any electric cylinder will be transmitted through coupling. The propagation amplification affects the output of all electric cylinders, leading to pose errors across the entire platform. Traditional single-position loop controllers... It is difficult to handle the coupled effects of external interference in a timely manner, thus limiting control performance.

[0035] The decoupling control strategy based on force sensor feedback proposed in this invention is based on the core idea of ​​constructing an inner-loop force control loop. This loop ensures that each electric cylinder operates independently at the force control level, while also being distinct from position space, generating compensation only for interfering forces, thereby controlling the controlled object. Diagonalize to, , This diagonalization achieves dynamic decoupling, suppresses interference propagation, and isolates the interference within a single electric cylinder, avoiding coupling amplification effects and thus significantly improving system stability. It simplifies control design and enhances response performance; essentially, the force sensor feedback directly measures the interference force and generates compensation commands in real time, eliminating the system's reliance on a precise dynamic model. Diagonalization ensures that compensation is applied only to local interference, making the system highly robust to parameter variations, unmodeled dynamics, and external disturbances.

[0036] In summary, this invention proposes a parallel platform anti-interference decoupling control system based on multi-sensor fusion. This system constructs a multi-source sensing network composed of three types of sensors: joint space motion quantities, joint space forces, and operational space inertial motion information. Based on this network, three functionally decoupled control loops are deployed in parallel within the controller, ultimately forming a high-performance control paradigm fundamentally different from traditional model-dependent paths. This system achieves "decoupling" and "directness" in anti-interference control. Specifically, the system ensures stable tracking of the macroscopic trajectory through the main tracking loop, establishing the control benchmark. Using an inertial feedback loop, and with the aid of a simplified inertial model and force Jacobian mapping, feedback compensation is provided for the system-dominant inertial dynamic effects, achieving initial decoupling at the dynamic level. Crucially, the force interference feedback loop constructs a local force closed loop by directly comparing the measured values ​​of joint space forces with the desired commands, enabling real-time sensing and direct cancellation of all residual interferences, including unmodeled dynamics, branch couplings, and external impacts. The coordinated operation of these three loops allows the system to effectively cope with multi-source interferences such as foundation vibration and load changes without relying on a precise and complex global nonlinear dynamic model.

[0037] It should be noted that all directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of the present invention are only used to explain the relative positional relationship and movement of each component in a certain specific posture (as shown in the figure). If the specific posture changes, the directional indication will also change accordingly.

[0038] Furthermore, in this invention, descriptions involving terms such as "first," "second," and "a" are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0039] In this invention, unless otherwise explicitly specified and limited, the terms "connection," "fixed," etc., should be interpreted broadly. For example, "fixed" can mean a fixed connection, a detachable connection, or an integral part; it can mean a mechanical connection or an electrical connection; it can mean a direct connection or an indirect connection through an intermediate medium; it can mean the internal communication of two components or the interaction between two components, unless otherwise explicitly limited. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0040] Furthermore, the technical solutions of the various embodiments of the present invention can be combined with each other, but only if they are feasible for those skilled in the art. If the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such combination of technical solutions does not exist and is not within the scope of protection claimed by the present invention.

Claims

1. A parallel platform anti-interference decoupling control system based on multi-sensor fusion, comprising a parallel mechanism, a multi-source sensing system, and a controller, wherein the parallel mechanism comprises a fixed platform, a moving platform, and multiple drive branches connecting the two, characterized in that: The multi-source sensing system includes: a first sensing unit disposed on each drive branch for acquiring the actual motion of the drive joint in the joint space; a second sensing unit disposed on each drive branch for detecting the actual output force of the drive joint in the joint space; and a third sensing unit disposed on the moving platform for measuring the inertial motion information in the operating space of the moving platform. By fusing information from multiple sensor systems, the controller is configured to execute three functionally decoupled control loops in parallel. It generates control commands to drive the parallel mechanism by fusing information from the multiple sensor systems. The control loops include: The main tracking loop obtains the expected motion of the driving joint based on the desired trajectory through inverse kinematics solution, and compares it with the actual motion to generate joint space error compensation. At the same time, it obtains the operation space error based on inertial motion information and maps it to the joint space compensation amount, and then superimposes it to achieve the reference output command for macroscopic motion tracking. The force disturbance feedback loop generates a compensating force to suppress unmodeled dynamics and external disturbances based on the deviation between the actual output force and the reference output force command. An inertial feedback loop, based on inertial motion information in the operating space, generates a feedback force to compensate for unmodeled inertial dynamics by mapping the inertial model to the force Jacobian and comparing it with the actual output force.

2. The anti-interference decoupling control system for a parallel platform based on multi-sensor fusion as described in claim 1, characterized in that, The first sensing unit is an encoder built into the drive motor of the drive joint, and the actual motion measured is the linear displacement and / or motion speed of the push rod at the corresponding joint.

3. The anti-interference decoupling control system for a parallel platform based on multi-sensor fusion as described in claim 1, characterized in that, The second sensing unit is a single-dimensional force sensor installed in series in the force flow path of the drive branch, and the central detection axis of the single-dimensional force sensor and the thrust axis of the drive branch are configured to remain collinear, and the actual output force detected is the output force along the joint axis.

4. The anti-interference decoupling control system for a parallel platform based on multi-sensor fusion as described in claim 3, characterized in that, The unidimensional force sensor is connected to the drive chain via a preload bolt.

5. The anti-interference decoupling control system for a parallel platform based on multi-sensor fusion as described in claim 3, characterized in that, The force interference feedback loop includes a low-pass filter, which filters the actual output force measured by the second sensing unit to suppress high-frequency bending moment interference generated by the swing of the drive branch.

6. The anti-interference decoupling control system for a parallel platform based on multi-sensor fusion as described in claim 1, characterized in that, The third sensing unit is an inertial measurement unit rigidly fixed to the moving platform, and the measurement axis of the inertial measurement unit is aligned with the working coordinate system of the moving platform. The measured spatial inertial motion information is the linear acceleration and angular acceleration of the moving platform.

7. The anti-interference decoupling control system for a parallel platform based on multi-sensor fusion as described in claim 6, characterized in that, The inertial model upon which the inertial feedback loop depends is a simplified model based on the mass and rotational inertia of the moving platform.

8. The anti-interference decoupling control system for a parallel platform based on multi-sensor fusion as described in claim 7, characterized in that, The feedback force in the inertial feedback loop is generated through the following steps: Based on the inertial motion information measured by the third sensing unit, and combined with the inertial matrix formed by the equivalent mass and moment of inertia of the moving platform, the inertial force in the operating space is calculated. By transposing the force Jacobian matrix of the parallel mechanism, the inertial force in the operating space is mapped to the joint space, and the difference between the inertial force and the actual output force is used to generate a feedback force to compensate for the unmodeled inertial dynamics.

9. The anti-interference decoupling control system for a parallel platform based on multi-sensor fusion as described in claim 1, characterized in that, The main tracking loop includes a proportional-derivative controller, which is configured to generate a reference output command based on the deviation between the expected and actual joint motion.

Images