Robotic anti-jam integrated circuit system based on multi-sensor fusion

Abstract

The application discloses a kind of robot interference immunity integrated circuit systems based on multi-sensor fusion, it is related to automation control technical field, it aims at solving the technical problem that workpiece geometric precision and surface integrity are difficult to guarantee in coordination in the processing of ultra-precision hole due to the complex multi-modal vibration of boring bar cannot be real-time decoupling and accurate compensation, including: sensor array, it is configured on the key structure of industrial robot end effector, for real-time acquisition and mechanical vibration related multidimensional physical signal;Special integrated circuit is connected with sensor array and robot controller signal, special integrated circuit includes: vibration reconstruction and compensation amount determination module, for receiving and fusion processing multidimensional physical signal collected by sensor array;The application realizes the identification of complex dynamic deformation of boring bar in processing by constructing real-time vibration deformation modal reconstruction and mapping system based on spatial distributed sensing array and special integrated circuit.

Description

Technical Field

[0001] This invention relates to the field of automation control technology, and more specifically, to a robot anti-interference integrated circuit system based on multi-sensor fusion. Background Technology

[0002] In the field of modern high-end equipment manufacturing, especially in industries such as aerospace and precision optics where the performance requirements for parts are extremely stringent, ultra-precision hole machining technology plays a crucial role. In the machining of micro-deep holes in aero-engine fuel nozzles, the hole diameter is extremely small, the depth-to-diameter ratio is large, and near-limit requirements are placed on the geometric accuracy of the hole wall (such as roundness and cylindricity) and surface integrity (no micro-cracks, controllable residual stress). The machining quality of such parts directly determines the performance, efficiency, and reliability of core equipment.

[0003] Currently, the mainstream process for achieving such ultra-high precision hole machining relies on precision boring or honing. However, a long-standing and difficult-to-overcome technical bottleneck severely restricts further improvements in machining accuracy: the unpredictable mechanical vibrations generated by the boring bar or honing bar during machining due to cutting forces, clamping, and its own dynamic characteristics. This vibration is not a simple overall swaying, but rather manifests as complex multimodal coupled bending and torsional deformation of the bar.

[0004] This leads to a serious problem: during machining, the actual trajectory of the tool deviates dynamically and slightly from its theoretical trajectory. This instantaneous deviation is "carved" onto the workpiece hole wall, forming specific shape errors (such as ellipses, triangular circles, etc.) and deteriorating the surface micro-morphology. Since vibration occurs in real time and varies with cutting conditions (such as tool wear and allowance changes), traditional offline measurement-compensation techniques are completely ineffective because the compensation amount cannot keep up with the changes in vibration state. Online monitoring systems based on a single sensor (such as a single accelerometer or spindle encoder) can only sense the overall vibration intensity and cannot decouple and reconstruct the root cause of specific shape errors—that is, the specific dynamic deformation mode of the boring bar in space—and therefore cannot generate targeted, high-precision compensation commands. This results in existing technologies constantly struggling to balance "geometric accuracy" and "surface integrity" when facing ultra-precision machining of micro-deep holes, making it difficult to improve yield and machining efficiency, becoming a key obstacle restricting the development of high-end manufacturing. In view of this, we propose a robot anti-interference integrated circuit system based on multi-sensor fusion. Summary of the Invention

[0005] The purpose of this invention is to provide a robot anti-interference integrated circuit system based on multi-sensor fusion to solve the technical problem that the complex multimodal vibration of the boring bar cannot be decoupled and accurately compensated in real time during ultra-precision hole machining, which makes it difficult to ensure the geometric accuracy and surface integrity of the workpiece in a coordinated manner.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a robot anti-interference integrated circuit system based on multi-sensor fusion, comprising: A sensor array, which is configured on a key structure of the end effector of an industrial robot, is used to collect multi-dimensional physical signals related to mechanical vibration in real time. Application-specific integrated circuit (ASIC), which is connected to the sensor array and the robot controller via signals, the ASIC comprising: The vibration reconstruction and compensation amount determination module is used to receive and fuse the multi-dimensional physical signals collected by the sensor array, reconstruct the real-time vibration deformation mode of the key structure, and determine the dynamic compensation amount to offset the trajectory or position error caused by vibration based on the synchronization relationship between the real-time vibration deformation mode and the motion phase. The dynamic compensation module is used to generate high-frequency compensation control commands based on the dynamic compensation amount; A micro-displacement actuator, integrated at the end of the end effector, receives compensation control commands output by the dynamic compensation module and executes actions to offset vibration errors in real time. The vibration reconstruction and compensation determination module is configured to perform iterative learning feedforward control, which records error information determined based on vibration deformation modes within one motion cycle, applies compensation based on the recorded error information in the subsequent corresponding phase, and continuously updates the compensation strategy.

[0007] This invention, by constructing a real-time vibration deformation mode reconstruction and mapping system based on a spatially distributed sensor array and a dedicated integrated circuit, achieves for the first time the identification of complex dynamic deformations of boring bars during machining, laying an irreplaceable sensing foundation for online precise compensation. This invention utilizes a sensor array, including fiber optic gratings, distributed along the boring bar to capture the microscopic strain field at multiple points in space. Through a modal decoupling algorithm embedded in the dedicated integrated circuit, the system can decompose and restore the mixed strain signal in real time into the instantaneous amplitude and phase of each independent vibration mode (such as first-order bending and torsion) of the boring bar. More importantly, through a preset deformation mapping relationship, it can accurately convert abstract modal coordinates into the real-time spatial position error of the tool tip within the machining plane. This progressive analytical process from mixed signals to independent modes and then to specific errors enables the system to accurately locate each vibration root cause leading to hole distortion, achieving a deep understanding of the interference source. This overcomes the limitation of existing technologies that cannot correlate vibration modes with specific machining errors, improving the accuracy of subsequent compensation control.

[0008] Preferably, the sensor array includes at least a plurality of fiber Bragg grating sensors distributed on the key structure, constituting a fiber Bragg grating sensor array for measuring spatial strain fields and acquiring multi-point spatial strain signals; the key structure is a boring bar or honing bar used in precision machining.

[0009] Preferably, the sensor array further includes one or more of the following sensors: An inertial measurement unit or accelerometer is installed at the base of the end effector or the root of the key structure to collect the foundation vibration acceleration signal; A force or torque sensor located near the spindle is used to indirectly monitor cutting force fluctuations; Acoustic emission sensors are used to collect signals of microscopic yielding or crack initiation in workpiece materials during processing.

[0010] Preferably, the vibration reconstruction and compensation determination module includes: A high-speed synchronous acquisition unit is used to synchronously acquire multi-dimensional physical signals from each sensor in the sensor array at a sampling rate higher than the target suppressed vibration frequency; The vibration mode decoupling unit is used to perform real-time calculation on the spatial multi-point strain signals collected by the fiber optic grating sensor array, and to separate and identify the instantaneous amplitude and phase information of at least the first-order bending vibration mode and the first-order torsional vibration mode of the key structure. The deformation mapping unit is used to map the decoupled vibration mode information to the machining tip or point of action of the end effector and calculate the real-time spatial position error caused by vibration.

[0011] Preferably, the vibration mode decoupling unit achieves mode separation by solving the transformation relationship from strain field to modal coordinates, specifically: ; In the formula, Indicates at discrete sampling time Located on the critical structure Strain values ​​at each measurement point; Indicates the first The first vibration mode in the 1st order Strain modal values ​​at each measurement point; Indicates at time No. Modal coordinates of the first vibration mode; This indicates the total number of modal orders considered; Indicates the first Each measurement point at time [time] Measurement noise.

[0012] Preferably, the vibration reconstruction and compensation determination module further includes an iterative learning control unit, which includes a dedicated data storage and a learning algorithm processor for performing iterative learning feedforward control; The learning algorithm processor is configured to: take each rotation of the spindle or robot joint or each repetitive motion cycle as a learning cycle; in the first learning cycle, record the relationship between the position error calculated by the deformation mapping unit and the motion phase to form a first cycle error mapping table; in the second and subsequent learning cycles, call the error mapping table stored in the previous cycle and generate feedforward compensation in advance at the corresponding motion phase point; at the same time, update the error mapping table according to the error measured in real time in the current cycle.

[0013] Preferably, the iterative learning feedforward control law implemented by the iterative learning control unit is: ; In the formula, and They represent the first time. and the The corresponding motion phase in each learning cycle The feedforward compensation control quantity; Indicates the first Phase of each learning cycle Real-time position error caused by vibration; Indicates phase The learning gain function at the location; Indicates the phase of motion; The index number represents the learning cycle.

[0014] Preferably, the iterative learning control unit is further configured to receive signals from the acoustic emission sensor; When a specific vibration mode is identified as being strongly correlated with an acoustic emission characteristic signal that characterizes surface quality deterioration, the compensation weight for the error component of that specific vibration mode is adaptively adjusted. Specifically, the adjusted learning law is as follows: ; In the formula, Indicates the first Each learning cycle, phase From the first The position error component caused by a specific vibration mode; Indicates phase Acoustic emission signal characteristic values ​​collected and analyzed from nearby locations; Indicates that for the first Additional reinforcement learning gain function for the hazard mode, when the acoustic emission characteristics are related to the mode When strongly correlated, this gain function is activated or increased, causing the control law to focus more on suppressing the mode.

[0015] Preferably, the dynamic compensation module includes: A digital compensation instruction generation unit is used to convert the dynamic compensation amount into digital instructions; An integrated high-frequency high-voltage drive unit is integrated on the same chip as the digital compensation instruction generation unit, which is used to convert the digital instruction into a high-voltage analog drive signal that can directly drive the micro-displacement actuator. The micro-displacement actuator is either a piezoelectric ceramic actuator or a magnetostrictive actuator.

[0016] A robot anti-interference control method based on multi-sensor fusion includes the following steps: S1: Real-time acquisition of multi-dimensional physical signals of key structures of the end effector of industrial robots through a distributed sensor array; S2: In the application-specific integrated circuit, the vibration deformation mode of the key structure in the processing plane is reconstructed in real time based on the dynamic strain signal in the multi-dimensional physical signal; S3: Based on the synchronization relationship between the vibration deformation mode and the motion phase, determine the dynamic compensation amount, wherein iterative learning feedforward control is executed to optimize the compensation amount for the current and future cycles using vibration error information from historical cycles; S4: The application-specific integrated circuit generates high-frequency compensation control commands to drive the micro-displacement actuator integrated at the end of the end effector to offset the trajectory or positional error caused by vibration in real time.

[0017] Compared with the prior art, the beneficial effects of the present invention are: 1. This invention, by constructing a real-time vibration deformation mode reconstruction and mapping system based on a spatially distributed sensor array and a dedicated integrated circuit, achieves for the first time the identification of complex dynamic deformations of boring bars during machining, laying an irreplaceable sensing foundation for online precise compensation. This invention utilizes a sensor array, including fiber optic gratings, distributed along the boring bar to capture the microscopic strain field at multiple points in space. Through a modal decoupling algorithm embedded in the dedicated integrated circuit, the system can decompose and restore the mixed strain signal in real time into the instantaneous amplitude and phase of each independent vibration mode (such as first-order bending and torsion) of the boring bar. More importantly, through a preset deformation mapping relationship, it can accurately convert abstract modal coordinates into the real-time spatial position error of the tool tip in the machining plane. This progressive analytical process from mixed signals to independent modes and then to specific errors enables the system to accurately locate every vibration source causing hole distortion, achieving a deep understanding of the interference source. This overcomes the limitation of existing technologies that cannot correlate vibration modes with specific machining errors, improving the accuracy of subsequent compensation control.

[0018] 2. This invention also integrates an intelligent compensation decision-making mechanism based on iterative learning feedforward control, endowing the system with self-evolution capabilities and effectively addressing the time-varying and periodic disturbances in the machining process. Simply sensing errors is insufficient for high-precision compensation; a control strategy with predictive and optimization capabilities is also required. The iterative learning control employed in this invention cleverly utilizes the inherent periodicity of the precision boring process (e.g., one revolution of the spindle). The system treats each motion cycle as a learning round, and the phase-error mapping relationship recorded in the previous cycle is used to generate the feedforward compensation amount for the corresponding phase in the next cycle. Through the high-speed computation of the dedicated integrated circuit, this learning process continuously iterates, enabling the compensation strategy to continuously self-optimize and asymptotically offset recurring periodic vibration errors. Compared to traditional fixed-parameter feedback control, this strategy does not rely on a precise, fixed controlled object model, thus exhibiting extremely strong robustness to slow time-varying factors such as tool wear and changes in working conditions. It allows the compensation system to proactively predict and act in advance, rather than passively responding, thereby achieving maximum suppression of periodic disturbances and significantly improving the long-term stability and consistency of machining.

[0019] 3. This invention also innovatively achieves a leap from geometric error compensation to shape-property synergistic control by integrating multi-source process signals such as acoustic emission, ensuring dimensional accuracy while actively maintaining the internal surface integrity of the workpiece. After solving the geometric accuracy problem, different vibration modes cause different levels of damage to the workpiece. Although certain specific vibration modes have little impact on roundness, they can easily cause microscopic tearing or crack initiation in the hole wall material, damaging surface integrity. This invention introduces an acoustic emission sensor to monitor the signals emitted by microscopic damage to the material during processing in real time. When the system identifies a specific vibration mode that is strongly correlated with the acoustic emission characteristic signal, it determines that the mode is a harmful mode. At this time, a dedicated learning algorithm adaptively adjusts the control strategy, significantly strengthening the suppression weight of the error component of that mode in the compensation command. This means that the system can not only compensate based on shape error, but also intelligently allocate control resources based on quality risk, prioritizing the elimination of the vibration source that is most harmful to the workpiece performance. This breaks through the limitations of traditional vibration control that focuses solely on dimensional accuracy, achieving coordinated assurance of the final comprehensive service performance of parts (i.e., geometry and surface physical properties), elevating machining quality control to a whole new level, and is of milestone significance for ensuring the reliability and lifespan of parts under extreme conditions such as aero engines. Attached Figure Description

[0020] Figure 1 This is a diagram of the overall system architecture of the present invention; Figure 2 This is a detailed architecture diagram of the internal modules of the application-specific integrated circuit of the present invention; Figure 3This is a flowchart of the overall anti-interference control method of the present invention; Figure 4 This is a detailed flowchart of the vibration deformation mode reconstruction and error calculation of the present invention; Figure 5 This is a flowchart of the iterative learning feedforward control execution process of the present invention. Detailed Implementation

[0021] like Figures 1 to 4 As shown, the present invention relates to a robot anti-interference integrated circuit system based on multi-sensor fusion, applied to the precision machining end effector of an industrial robot, for suppressing mechanical vibration interference during machining, comprising: A sensor array, configured on a key structure of the end effector, is used to acquire multi-dimensional physical signals related to mechanical vibration in real time. The multi-dimensional physical signals include at least the dynamic strain signals of the key structure at multiple points in space. The key structure is a boring bar or honing bar used in precision boring or honing, and the end effector is a machine tool spindle.

[0022] In another embodiment of the present invention, the sensor array includes a plurality of fiber Bragg grating sensors embedded or attached to the interior or surface of the boring bar or honing bar. The plurality of fiber Bragg grating sensors are distributed along the axial and circumferential directions of the boring bar or honing bar to form a fiber Bragg grating sensor array for measuring the spatial strain field and acquiring multi-point strain signals in space. The fiber Bragg grating sensor array is configured to simultaneously measure the bending strain and torsional strain of the boring bar or honing bar.

[0023] In another embodiment of the present invention, the sensor array further includes one or more of the following sensors: An inertial measurement unit or accelerometer is installed at the base of the end effector or the root of the key structure to collect the foundation vibration acceleration signal; A force or torque sensor located near the spindle is used to indirectly monitor cutting force fluctuations; Acoustic emission sensors are used to collect signals of microscopic yielding or crack initiation in workpiece materials during processing.

[0024] Application-specific integrated circuit (ASIC), which is connected to the sensor array and the robot controller via signals, the ASIC comprising: The vibration reconstruction and compensation determination module is used to receive and process the multi-dimensional physical signals collected by the sensor array in real time, reconstruct the real-time vibration deformation mode of the key structure in the machining plane based on the dynamic strain signal, and determine the dynamic compensation amount to offset the trajectory or position error caused by vibration according to the synchronization relationship between the real-time vibration deformation mode and the spindle rotation phase. In another embodiment of the present invention, the vibration reconstruction and compensation amount determination module includes: A high-speed synchronous acquisition unit is used to synchronously acquire multi-dimensional physical signals from each sensor in the sensor array at a sampling rate higher than the target suppression vibration frequency. The vibration mode decoupling unit, which is implemented based on hardware logic circuits or a dedicated processor core, is used to perform real-time calculation on the spatial multi-point strain signals collected by the fiber optic grating sensor array, and to separate and identify the instantaneous amplitude and phase information of at least the first-order bending vibration mode and the first-order torsional vibration mode of the key structure. The deformation mapping unit is used to map the decoupled vibration mode information to the machining tip or point of action of the end effector and calculate the real-time spatial position error caused by vibration.

[0025] The vibration mode decoupling unit achieves mode separation by solving the transformation relationship from strain field to modal coordinates. Specifically, at any sampling time... The key structure on the first Strain at each measurement point With each modal coordinate The relationship is described by the following formula: ; In the formula: Indicates at discrete sampling time Located on the critical structure The strain value directly measured by a sensor (such as a fiber Bragg grating) at each measurement point. It is the raw input for physical measurement; Indicates the first The first vibration mode in the 1st order The strain modal values ​​at each measurement point are constants that are pre-calibrated through experimental modal analysis or finite element analysis. These constitute the system's knowledge base and describe the unique strain distribution pattern of each mode in space. Indicates at time No. The modal coordinates of each vibration mode are the core output of the algorithm. Their magnitude and variation directly represent the instantaneous amplitude of the vibration at that mode. The vibration mode decoupling unit obtains the modal coordinates of each mode by solving the above equations in real time or by applying a pre-stored decoupling matrix. ; This indicates the total order of the vibration modes that the system considers and intends to separate (e.g., first-order bending, second-order bending, first-order torsion, etc.). Indicates the first Each measurement point at time [time] Measurement noise; Explanation of the operational logic: This formula describes the mathematical transformation process from physical measurement signals to abstract vibration modes. The system obtains the inherent strain mode matrix of the structure through pre-calibration. This is a constant matrix reflecting the vibration characteristics of the structure itself. During actual operation, a dedicated integrated circuit acquires strain values ​​from multiple sensor measurement points in real time. It is regarded as a vibration of various modes. According to the contribution of vibration mode ( The linear superposition of the strain signals and the instantaneous vibration intensity (i.e., modal coordinates) of each independent mode can be decoupled from the mixed strain signals in real time by solving this system of linear equations or by applying a pre-calculated inverse matrix. This process essentially decomposes the complex, coupled spatial strain field into independent modal components that are easy to understand and control. Through the above calculations, the system achieves real-time and precise decoupling of complex coupled vibrations in modal space. This allows the control system to independently identify, analyze, and selectively suppress each harmful vibration, much like adjusting the volume of different audio tracks, rather than processing the mixed vibration signal indiscriminately. This lays a fundamental foundation for subsequent high-precision, targeted compensation, representing a crucial leap from perceiving vibration to understanding it.

[0026] Specifically, the deformation mapping unit will decouple the modal coordinates The mapping relationship to the tool tip position error is as follows: ; In the formula, For the first The mapping function from the first mode to the tool tip position error has the dimension of length per generalized coordinate (e.g., micrometer / modal unit, determined through offline calibration experiments). It represents generalized displacement, and its unit can be normalized by modal mass or proportional to physical displacement (micrometer); Indicates phase The corresponding discrete sampling time.

[0027] The real-time position error The result is obtained by summing the error components of all modes: ,in Corresponding to the Error components caused by first mode; In another embodiment of the present invention, the vibration reconstruction and compensation determination module further includes: An iterative learning control unit, comprising a dedicated data storage and a learning algorithm processor, is used to execute the iterative learning feedforward control; The learning algorithm processor is configured to: take each rotation of the spindle or robot joint or each repetitive motion cycle as a learning cycle; record the relationship between the position error calculated by the deformation mapping unit and the motion phase in the first learning cycle to form a first cycle error mapping table; in the second and subsequent learning cycles, call the error mapping table stored in the previous cycle to generate feedforward compensation in advance at the corresponding motion phase point; at the same time, update the error mapping table according to the error measured in real time in the current cycle to optimize the compensation amount in subsequent cycles.

[0028] The iterative learning feedforward control law implemented by the iterative learning control unit is described by the following formula: ; In the formula: and They represent the first time. and the In each learning cycle, corresponding to the motion phase The feedforward compensation control quantity is the target value of the instruction output to the micro-displacement actuator. The units of both are volts (V) or micrometers (μm). Indicates the first Phase of each learning cycle The real-time position error caused by vibration is derived from the spatial position deviation mapped from the calculated modal coordinates. This error serves as the basis for learning and is measured in micrometers (μm). Indicates phase The learning gain function is a key parameter that determines the learning speed and stability. It can be a constant or a function that varies with the phase. It is used to adjust the degree of influence of error information on the update of the compensation amount. The dimensions are configured to make the product With the feedforward compensation control quantity They have the same physical dimensions; It represents the motion phase, which in rotary machining is usually the spindle rotation angle, and characterizes the progress within one motion cycle; The index number representing the learning cycle increases with each machining cycle (e.g., one revolution of the spindle); Operational logic explanation: This formula defines an advanced feedforward control law with self-learning capabilities. Its core idea is to utilize the periodicity of the processing procedure to... (The sentence is incomplete and requires more context to translate accurately.) Vibration error observed within ) After a specific learning gain After adjustment, the increased amount will be added to the current period as compensation. The amount of compensation used Up, thus forming the basis for the next cycle ( A more optimized feedforward compensation amount This is an iterative, self-optimizing process: errors drive the updating of compensation amounts, enabling the system to proactively predict and counteract periodically recurring disturbances, rather than passively responding. This iterative learning control law endows the system with the ability to learn from experience. It does not rely on precise, fixed mathematical models, but rather continuously self-corrects and optimizes through periodic practice. As the learning cycle increases, the feedforward compensation for periodic disturbances becomes increasingly accurate, theoretically allowing the tracking error to asymptotically converge to zero. This significantly improves the system's ultimate ability to suppress recurring disturbances and provides adaptability to slowly changing system characteristics (such as tool wear), achieving intelligent precision maintenance.

[0029] In another embodiment of the present invention, the iterative learning control unit is further configured to receive signals from the acoustic emission sensor and adaptively adjust the compensation weight for the specific vibration mode when a strong correlation is identified between a specific vibration mode and an acoustic emission characteristic signal characterizing surface quality deterioration.

[0030] Specifically, the real-time position error Composed of contributions from different vibration modes, denoted as ,in Corresponding to the Error components caused by first mode; when the acoustic emission signal characteristics Exceeding the threshold and with a specific modality coordinates When statistically strong correlation occurs, the learning algorithm processor learns the gain function. Adjustments were made to enhance the modal error. The inhibition, the adjusted learning law is: ; In the formula: Indicates the first Each learning cycle, phase At this point, by the first The position error component caused by a specific vibration mode of order. It is composed of the coordinates of that mode. It is calculated through a fixed mapping relationship; Indicates phase Acoustic emission signal feature values ​​collected and analyzed nearby are used to characterize the risk of deterioration of the processed surface quality (such as micro-tears and crack initiation). Indicates that for the first Additional reinforcement learning gain function for the hazard mode. When acoustic emission characteristics are related to the mode... When strongly correlated, this gain function is activated or increased, causing the control law to focus more on suppressing that mode. and All are dimensionless proportionality coefficients; Explanation of the operational logic: This formula introduces a reinforcement learning term based on multi-sensor information fusion on top of the basic iterative learning law. When the system senses a specific vibration mode (e.g., through acoustic emission sensors)... This is causing harmful process effects (such as surface micro-damage, manifested as) When the threshold is exceeded, the algorithm no longer treats all modal errors equally. It will target the specific modal error component that is identified as harmful. Apply an additional, reinforcing learning gain. This allows for a higher weighting and more aggressive suppression of this component in the update of the total compensation. This achieves an intelligent strategy upgrade from suppressing all vibrations to prioritizing the suppression of the most harmful ones. This formula embodies the highest level of shape-based collaborative control. The system not only focuses on geometric accuracy (suppression error) Furthermore, it indirectly senses the surface integrity quality of the workpiece through acoustic emission signals. When a specific vibration mode is detected as damaging the intrinsic quality of the workpiece, the system can dynamically adjust the control priority and implement precise surgical suppression. This breaks through the limitations of traditional vibration control, which only aims to ensure dimensional accuracy, and achieves synergistic protection of the overall performance (geometric accuracy and surface integrity) of the final part, greatly improving the overall quality and yield of the processing technology.

[0031] Among them, it is clear , , Relationship: Let the spindle rotation period be The discrete sampling interval is Then phase With time Relationship: ; Discrete sampling time Corresponding time: ; Learning cycle Corresponding to the One complete rotation cycle: ; The dynamic compensation module is used to generate high-frequency compensation control commands based on the dynamic compensation amount; A micro-displacement actuator, integrated at the end of the end effector, receives compensation control commands output by the dynamic compensation module and executes actions to offset vibration errors in real time. In another embodiment of the present invention, the dynamic compensation module includes: A digital compensation instruction generation unit is used to convert the dynamic compensation amount determined by the vibration reconstruction and compensation amount determination module into digital instructions; An integrated high-frequency high-voltage drive unit is integrated on the same chip as the digital compensation instruction generation unit, which is used to convert the digital instructions into high-voltage analog drive signals that can directly drive the micro-displacement actuator; wherein, the micro-displacement actuator is a piezoelectric ceramic actuator or a magnetostrictive actuator.

[0032] The integrated high-frequency high-voltage drive unit includes an on-chip digital isolation circuit and a high-voltage power amplifier circuit, which can realize low-delay signal conversion and power output from the low-voltage digital control domain to the high-voltage analog drive domain.

[0033] The application-specific integrated circuit is configured to perform iterative learning feedforward control, which records vibration error information within one processing cycle, applies compensation based on the recorded error information in the subsequent corresponding phase, and continuously updates the compensation strategy.

[0034] In another embodiment of the present invention, the application-specific integrated circuit is a system-on-a-chip that integrates a high-speed analog front-end, a digital signal processor, a hardware logic acceleration unit, and a mixed-signal driving unit.

[0035] The robot anti-interference integrated circuit system is applied to the micro-deep hole machining scenario. The end effector is a micro boring bar, the machining motion is ultra-precision boring, and the micro-deep hole machining scenario is specifically the machining of micro-deep holes for aero-engine fuel nozzles.

[0036] Example 2: A robot anti-interference control method based on multi-sensor fusion, applied to a robot anti-interference integrated circuit system, the method includes the following steps: S1: Real-time acquisition of multi-dimensional physical signals of key structures of the end effector of industrial robots through a distributed sensor array; S2: In the application-specific integrated circuit, the vibration deformation mode of the key structure in the processing plane is reconstructed in real time based on the dynamic strain signal in the multi-dimensional physical signal; In another embodiment of the present invention, the real-time reconstruction of vibration deformation modes in step S2 specifically includes: real-time decoupling of the spatial multi-point strain signals collected by the fiber optic grating sensor array to separate the bending vibration mode and torsional vibration mode of the key structure.

[0037] S3: Based on the synchronization relationship between the vibration deformation mode and the motion phase, determine the dynamic compensation amount, wherein the dedicated integrated circuit performs iterative learning feedforward control and optimizes the compensation amount for the current and future cycles using vibration error information from historical cycles; In another embodiment of the present invention, the iterative learning feedforward control performed in step S3 specifically includes: S301: One learning cycle is defined as one revolution of the spindle. S302: In the first learning cycle, measure and record the mapping relationship between the error caused by vibration and the spindle phase; S303: In the second learning cycle, based on the mapping relationship recorded in the first learning cycle, a feedforward compensation amount is generated and applied at the corresponding phase point; S304: In each subsequent learning cycle, the mapping relationship is updated using the error information from the previous cycle and used for feedforward compensation in the next cycle.

[0038] S4: The application-specific integrated circuit generates high-frequency compensation control commands to drive the micro-displacement actuator integrated at the end of the end effector to offset the trajectory or positional error caused by vibration in real time.

[0039] As another embodiment of the present invention, the method further includes: Acquire acoustic emission signals during the processing; The characteristics of the acoustic emission signal are correlated with the reconstructed vibration modes; When a specific vibration mode is determined to pose a threat to the integrity of the machined surface, the suppression priority and compensation weight of that specific vibration mode are increased.

[0040] Example 3: An industrial robot includes a robot body, a controller, and a robot anti-interference integrated circuit system based on multi-sensor fusion. The robot anti-interference integrated circuit system is communicatively connected to the controller and is used to improve the robot's anti-vibration interference capability and machining accuracy when performing precision machining tasks.

[0041] In an embodiment of the present invention, As another embodiment of the present invention, Working principle: This embodiment provides a robot anti-interference integrated circuit system based on multi-sensor fusion. When in use... The embodiments disclosed in this invention are preferred embodiments, but are not limited thereto. Those skilled in the art can easily understand the spirit of this invention based on the above embodiments and make different extensions and variations, but as long as they do not depart from the spirit of this invention, they are all within the protection scope of this invention.

Claims

1. A robot anti-interference integrated circuit system based on multi-sensor fusion, characterized in that, include: A sensor array, configured on a key structure of an industrial robot end effector, is used to acquire multi-dimensional physical signals related to mechanical vibration in real time. The sensor array includes at least a plurality of fiber Bragg grating sensors distributed on the key structure, forming a fiber Bragg grating sensor array for measuring spatial strain fields and acquiring multi-point spatial strain signals. Application-specific integrated circuit (ASIC), which is connected to the sensor array and the robot controller via signals, the ASIC comprising: The vibration reconstruction and compensation amount determination module is used to receive and fuse the multi-dimensional physical signals collected by the sensor array, reconstruct the real-time vibration deformation mode of the key structure, and determine the dynamic compensation amount to offset the trajectory or position error caused by vibration based on the synchronization relationship between the real-time vibration deformation mode and the motion phase. The dynamic compensation module is used to generate high-frequency compensation control commands based on the dynamic compensation amount; A micro-displacement actuator, integrated at the end of the end effector, receives compensation control commands output by the dynamic compensation module and executes actions to offset vibration errors in real time. The vibration reconstruction and compensation determination module is configured to perform iterative learning feedforward control, which records error information based on vibration deformation mode determination within one motion cycle, applies compensation based on the recorded error information on the subsequent corresponding phase, and continuously updates the compensation strategy. The vibration reconstruction and compensation determination module includes: A high-speed synchronous acquisition unit is used to synchronously acquire multi-dimensional physical signals from each sensor in the sensor array at a sampling rate higher than the target suppressed vibration frequency; The vibration mode decoupling unit is used to perform real-time calculation on the spatial multi-point strain signals collected by the fiber optic grating sensor array, and to separate and identify the instantaneous amplitude and phase information of at least the first-order bending vibration mode and the first-order torsional vibration mode of the key structure. The deformation mapping unit is used to map the decoupled vibration mode information to the machining tip or point of action of the end effector and calculate the real-time spatial position error caused by vibration. The dynamic compensation module includes: A digital compensation instruction generation unit is used to convert the dynamic compensation amount into digital instructions; An integrated high-frequency high-voltage drive unit is integrated on the same chip as the digital compensation instruction generation unit, which is used to convert the digital instructions into high-voltage analog drive signals that can directly drive the micro-displacement actuator.

2. The robot anti-interference integrated circuit system based on multi-sensor fusion according to claim 1, characterized in that, The key structure is a boring bar or honing bar used in precision machining.

3. The robot anti-interference integrated circuit system based on multi-sensor fusion according to claim 1, characterized in that, The sensor array also includes one or more of the following sensors: An inertial measurement unit or accelerometer is installed at the base of the end effector or the root of the key structure to collect the foundation vibration acceleration signal; A force or torque sensor located near the spindle is used to indirectly monitor cutting force fluctuations; Acoustic emission sensors are used to collect signals of microscopic yielding or crack initiation in workpiece materials during processing.

4. The robot anti-interference integrated circuit system based on multi-sensor fusion according to claim 2, characterized in that, The vibration mode decoupling unit achieves mode separation by solving the transformation relationship from strain field to modal coordinates. The specific relationship is as follows: ； In the formula, Indicates at discrete sampling time Located on the critical structure Strain values ​​at each measurement point; Indicates the first The first vibration mode in the 1st order Strain modal values ​​at each measurement point; Indicates at time No. Modal coordinates of the first vibration mode; This indicates the total number of modal orders considered; Indicates the first Each measurement point at time [time] Measurement noise.

5. A robot anti-interference integrated circuit system based on multi-sensor fusion according to claim 2, characterized in that, The vibration reconstruction and compensation determination module also includes an iterative learning control unit, which contains a dedicated data storage and a learning algorithm processor for performing iterative learning feedforward control. The learning algorithm processor is configured to: take each rotation of the spindle or robot joint or each repetitive motion cycle as a learning cycle; in the first learning cycle, record the relationship between the position error calculated by the deformation mapping unit and the motion phase to form a first cycle error mapping table; in the second and subsequent learning cycles, call the error mapping table stored in the previous cycle and generate feedforward compensation in advance at the corresponding motion phase point; at the same time, update the error mapping table according to the error measured in real time in the current cycle.

6. A robot anti-interference integrated circuit system based on multi-sensor fusion according to claim 5, characterized in that, The iterative learning feedforward control law implemented by the iterative learning control unit is as follows: ； In the formula, and They represent the first time. and the The corresponding motion phase in each learning cycle The feedforward compensation control quantity; Indicates the first Phase of each learning cycle Real-time position error caused by vibration; Indicates phase The learning gain function at the location; Indicates the phase of motion; The index number represents the learning cycle.

7. A robot anti-interference integrated circuit system based on multi-sensor fusion according to claim 6, characterized in that, The iterative learning control unit is also configured to receive signals from the acoustic emission sensor; When a specific vibration mode is identified as being strongly correlated with an acoustic emission characteristic signal that characterizes surface quality deterioration, the compensation weight for the error component of that specific vibration mode is adaptively adjusted. Specifically, the adjusted learning law is as follows: ； In the formula, Indicates the first Each learning cycle, phase From the first The position error component caused by a specific vibration mode; Indicates phase Acoustic emission signal characteristic values ​​collected and analyzed from nearby locations; Indicates that for the first Additional reinforcement learning gain function for the hazard mode, when the acoustic emission characteristics are related to the mode When strongly correlated, this gain function is activated or increased, causing the control law to focus more on suppressing the mode.

8. A robot anti-interference integrated circuit system based on multi-sensor fusion according to claim 1, characterized in that, The micro-displacement actuator is a piezoelectric ceramic actuator or a magnetostrictive actuator.

9. A robot anti-interference control method based on multi-sensor fusion, applied to the robot anti-interference integrated circuit system as described in any one of claims 1-8, characterized in that, Includes the following steps: S1: Real-time acquisition of multi-dimensional physical signals of key structures of the end effector of industrial robots through a distributed sensor array; S2: In the application-specific integrated circuit, the vibration deformation mode of the key structure in the processing plane is reconstructed in real time based on the dynamic strain signal in the multi-dimensional physical signal; S3: Based on the synchronization relationship between the vibration deformation mode and the motion phase, determine the dynamic compensation amount, wherein iterative learning feedforward control is executed to optimize the compensation amount for the current and future cycles using vibration error information from historical cycles; S4: The application-specific integrated circuit generates high-frequency compensation control commands to drive the micro-displacement actuator integrated at the end of the end effector to offset the trajectory or positional error caused by vibration in real time.

Images