Precise part self-adaptive machining system and method based on multi-modal perception

Abstract

The invention discloses a precise part self-adaptive machining system and method based on multi-modal sensing. The system integrates a multi-modal sensing module, a composite vibration control unit, a data processing and fusion module, a self-adaptive control module and a machining execution module. The multi-mode sensing module monitors various parameters in the machining process in real time through visual sense, force sense, acoustics, temperature and piezoelectric detection elements. And the data processing and fusion module adopts processing technologies such as wavelet transform and a deep learning algorithm to generate a processing state comprehensive evaluation vector. And the self-adaptive control module adjusts machining parameters in real time through composite vibration control, dynamic parameter optimization and a closed-loop feedback mechanism according to the evaluation result. The machining execution module is integrated with a multi-mode resonance ultrasonic vibration fast tool, and high-flexibility micro / nano structure machining is achieved. According to the method, the machining precision and efficiency are improved, the influences of tool abrasion and thermal errors are reduced, and the wide application prospect is achieved.

Description

Technical Field

[0001] The present invention belongs to the cross - field of intelligent manufacturing of precision parts and ultra - precision machining technology, and particularly relates to a precision part adaptive machining system and method based on multi - modal perception. Background Technique

[0002] Traditional precision machining systems rely on single - modal perception (such as only vision or force sense), and it is difficult to comprehensively capture the dynamic changes in the machining process (such as material deformation, tool wear, thermal error, vibration modal instability, etc.). Although the existing vibration - assisted machining technologies (such as Fast Tool Servo - FTS) can improve machining flexibility, they are limited by the insufficient bandwidth of a single vibration mode (usually below 3kHz) and cannot meet the requirements of ultra - precision manufacturing of high - complexity micro / nano structures. In addition, the existing technologies lack the deep integration of multi - modal perception and high - frequency vibration control, resulting in a lag in machining parameter adjustment and making it difficult to achieve real - time closed - loop optimization. Summary of the Invention

[0003] The purpose of the present invention is to provide a precision part adaptive machining system and method based on multi - modal perception to solve the problems raised in the above - mentioned background technique.

[0004] To achieve the above purpose, the present invention provides the following technical solution: A precision part adaptive machining system based on multi - modal perception, including: a multi - modal perception module, a composite vibration control unit, a data processing and fusion module, an adaptive control module, and a machining execution module;

[0005] The multi - modal perception module includes:

[0006] A vision sensor for real - time monitoring of the part surface topography, tool wear, and machining trajectory;

[0007] A force sensor integrated on the spindle to collect cutting force and vibration spectrum;

[0008] An acoustic sensor for capturing machining noise and acoustic emission signals to identify abnormal cutting states;

[0009] A temperature sensor for monitoring the thermal deformation of the machine tool and the temperature distribution in the machining area;

[0010] A piezoelectric detection element embedded in the ultrasonic vibration quick - tool to convert the vibration deformation into a high - frequency current signal in real - time and feedback the amplitude and phase difference of the vibration mode;

[0011] The data processing and fusion module uses wavelet transform and deep - learning algorithms to extract features from visual images, cutting - force spectra, and acoustic emission signals; based on Kalman filtering and inverse Fourier transform, it fuses multi - modal data and vibration current signals to generate a comprehensive evaluation vector of the machining state;

[0012] The adaptive control module includes:

[0013] Compound vibration control unit: It outputs two high-frequency sine signals (fundamental frequency and third harmonic frequency) with adjustable phases through an ultrasonic power supply, and drives a resonant ultrasonic quick tool through an adder circuit and a power amplifier to generate target trajectories such as sine waves and triangular waves.

[0014] Dynamic parameter optimization unit: Combining a reinforcement learning algorithm, it adjusts the vibration frequency, phase difference, cutting speed, and feed rate in real time according to multimodal feedback.

[0015] Closed-loop feedback unit: Based on the current signal of a piezoelectric detection element, it stabilizes the resonant frequency through frequency sweeping and impedance matching techniques to suppress the fluctuation of the processing load.

[0016] The processing execution module integrates a multimodal resonant ultrasonic vibration quick tool (including a diamond tool and an ultrasonic transducer), supports high-frequency (above 20 kHz) axial intermittent movement, and realizes high-flexibility micro / nano structure processing.

[0017] Furthermore, the compound vibration control unit includes an ultrasonic power supply, an adder circuit, and a power amplifier, which are used to generate a fundamental frequency and third harmonic frequency composite excitation signal to drive the resonant ultrasonic quick tool to generate multimodal axial vibration.

[0018] Furthermore, the adaptive control module integrates a closed-loop feedback mechanism of a piezoelectric detection element to stabilize the resonant frequency through frequency sweeping and impedance matching techniques.

[0019] Furthermore, an inverse Fourier transform and a reinforcement learning algorithm are adopted to realize the collaborative optimization of the vibration trajectory and cutting parameters.

[0020] A precision part adaptive processing method based on multimodal perception includes the following steps:

[0021] S1, Multimodal data synchronous acquisition: Real-time acquisition of visual, tactile, acoustic, temperature data, and vibration current signals;

[0022] S2, High-frequency vibration trajectory generation: Adjust the superposition of two frequency-doubled signals through an ultrasonic power supply to drive the diamond tool to generate a target waveform trajectory (such as a trapezoidal wave or a square wave);

[0023] S3, Dynamic evaluation of the processing state: Fusing multimodal features such as vibration phase difference, cutting force spectrum, and surface roughness to judge the tool wear level and thermal error amount;

[0024] S4, Closed-loop parameter optimization: According to the evaluation results, adjust the amplitude-frequency characteristics of the vibration mode and cutting parameters, and calibrate the resonant frequency in real time through piezoelectric detection feedback;

[0025] S5. Iterative Learning and Process Updating: Optimize the deep learning model based on historical processing data to improve the generalization ability of adaptive control

[0026] Technical Effects and Advantages of the Present Invention:

[0027] 1. Through the multi-source data fusion of visual, tactile, acoustic, temperature, and piezoelectric detection elements, combined with the dynamic adjustment of the high-frequency resonant vibration trajectory, global state perception and millisecond-level response of the processing process are achieved, solving the problem of accuracy loss caused by the lag of single-modal perception in traditional systems.

[0028] 2. Adopt the composite excitation of the first-order and third-order longitudinal vibration modes, support the generation of various trajectories such as sine waves, trapezoidal waves, and square waves, significantly improving the processing flexibility of complex micro / nano structures; at the same time, the working frequency reaches above 20 kHz, and the efficiency is increased by 6 times compared with the traditional FTS technology (3 kHz).

[0029] 3. Based on the real-time current signal of the piezoelectric detection element, combined with the sweep frequency technology and the reinforcement learning algorithm, dynamic calibration of the resonant frequency and impedance matching are achieved, effectively suppressing the fluctuation of the processing load, ensuring the stability of the vibration trajectory, and the processing accuracy reaches the sub-micron level.

[0030] 4. Through the optimization of the deep learning model driven by historical data, the system can adapt to different material and process requirements, reduce the dependence on manual parameter adjustment, lower production costs, and is applicable to high-value-added fields such as aerospace and optical devices.

[0031] 5. Real-time anomaly detection and parameter compensation of multi-modal data reduce tool abnormal wear and machine tool thermal error, extend the tool life by more than 25%, and reduce the comprehensive production cost by 20%. Description of the Drawings

[0032] Figure 1 is the system block diagram of the present invention;

[0033] Figure 2 is the flow chart of the present invention. Detailed Embodiments

[0034] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0035] As Figure 1 - Figure 2 shown, the present invention provides a specific embodiment of a precision part adaptive processing system and method based on multi-modal perception, which will be described in detail below in conjunction with the drawings and technical solutions.

[0036] System Composition

[0037] The precision part adaptive machining system based on multi-modal perception of the present invention mainly includes the following modules: a multi-modal perception module, a composite vibration control unit, a data processing and fusion module, an adaptive control module, and a machining execution module:

[0038] Multi-modal Perception Module

[0039] Vision sensor: Installed near the machining area, it is used to monitor the surface topography of the part, tool wear condition, and machining trajectory in real time to ensure good monitoring of machining accuracy and tool status;

[0040] Force sensor: Integrated on the machine tool spindle, it is used to collect the cutting force and vibration spectrum during the cutting process to help identify unstable states during the cutting process;

[0041] Acoustic sensor: Arranged in the machining area, it captures machining noise and acoustic emission signals to identify abnormal cutting states, such as tool breakage or internal defects in the material;

[0042] Temperature sensor: Distributed in key parts of the machine tool and the machining area, it is used to monitor the thermal deformation of the machine tool and the temperature distribution in the machining area to prevent the influence of thermal error on machining accuracy;

[0043] Piezoelectric detection element: Embedded in the ultrasonic vibration quick tool, it converts the vibration deformation into a high-frequency current signal in real time to provide information on the amplitude and phase difference of the vibration mode;

[0044] Composite Vibration Control Unit

[0045] It consists of an ultrasonic power supply, an adder circuit, and a power amplifier, and is used to generate a composite excitation signal of fundamental frequency and third harmonic frequency to drive the resonant ultrasonic quick tool to generate multi-modal axial vibration to meet the machining requirements of complex micro / nano structures;

[0046] Data Processing and Fusion Module

[0047] Use wavelet transform and deep learning algorithms to extract features from visual images, cutting force spectra, and acoustic emission signals;

[0048] Based on Kalman filtering and inverse Fourier transform, fuse multi-modal data and vibration current signals to generate a comprehensive evaluation vector of the machining state to provide data support for adaptive control;

[0049] Adaptive Control Module

[0050] Composite vibration control unit: Generate the required vibration trajectory by regulating the phase and amplitude of the high-frequency sine signal;

[0051] Dynamic parameter optimization unit: Combining with reinforcement learning algorithms, it adjusts vibration frequency, phase difference, cutting speed, and feed rate in real time according to multi-modal perception data to achieve dynamic optimization of machining parameters;

[0052] Closed-loop feedback unit: Based on the current signal of piezoelectric detection elements, it stabilizes the resonance frequency through frequency sweeping and impedance matching techniques, effectively suppressing load fluctuations during the machining process;

[0053] Machining execution module

[0054] Integrated multi-modal resonant ultrasonic vibration quick tool, supporting high-frequency (above 20 kHz) axial intermittent motion to achieve precision machining of high-flexibility micro / nano structures.

[0055] The machining method of the present invention includes the following steps:

[0056] Multi-modal data synchronous acquisition: Real-time acquisition of vision, force sense, acoustics, temperature data, and vibration current signals provides comprehensive data support for subsequent machining state evaluation and control;

[0057] High-frequency vibration trajectory generation: By adjusting the superposition of two frequency-doubled signals through an ultrasonic power supply, it drives the diamond tool to generate target waveform trajectories, such as sine waves, triangular waves, trapezoidal waves, or square waves, to meet different machining requirements;

[0058] Machining state dynamic evaluation: Integrating multi-modal features such as vibration phase difference, cutting force spectrum, and surface roughness, it judges the tool wear level and thermal error amount through a deep learning model to ensure the stability and accuracy of the machining process;

[0059] Closed-loop parameter optimization: According to the evaluation results, it adjusts the amplitude-frequency characteristics of the vibration mode and cutting parameters, such as vibration frequency, phase difference, cutting speed, and feed rate, and calibrates the resonance frequency in real time through piezoelectric detection feedback to achieve real-time optimization of machining parameters;

[0060] Iterative learning and process update: Continuously optimize the deep learning model based on historical machining data to improve the generalization ability of adaptive control and provide more accurate control strategies for future machining tasks.

[0061] Through the implementation of the present invention, it is possible to achieve comprehensive perception and adaptive control of the machining process of precision parts, significantly improve machining accuracy and efficiency, and at the same time reduce the impact of tool wear and thermal error on machining quality. This method and technical solution have broad application prospects in the fields of intelligent manufacturing and ultra-precision machining.

[0062] The applicant further declares that the present invention uses the above embodiments to illustrate the implementation method and device structure of the present invention, but the present invention is not limited to the above implementation manners, that is, it does not mean that the present invention must rely on the above methods and structures to be implemented. Those skilled in the art should understand that any improvement to the present invention, equivalent replacement of the implementation method selected by the present invention, addition of steps, selection of specific manners, etc., all fall within the protection scope and the disclosed scope of the present invention.

[0063] The present invention is not limited to the above implementation manners. All manners that adopt a structure similar to that of the present invention and its methods to achieve the purpose of the present invention are within the protection scope of the present invention.

Claims

1. Adaptive processing system for precision parts based on multimodal perception, characterized by: include: Multimodal perception module, composite vibration control unit, data processing and fusion module, adaptive control module and processing execution module; The multimodal perception module comprises: Visual sensors are used to monitor part surface topography, tool wear and processing trajectory in real time; Force sensor, used to be integrated into the spindle to collect cutting force and vibration spectrum; Acoustic sensor: used to capture processing noise and acoustic emission signals and identify abnormal cutting conditions; Temperature sensor: used to monitor the thermal deformation of machine tools and the temperature distribution in the processing area; Piezoelectric detection element: embedded with ultrasonic vibration knife, it converts vibration deformation into high-frequency current signal in real time and feeds back the amplitude and phase difference of vibration mode; The data processing and fusion module uses wavelet transform and deep learning algorithm to extract features from visual images, cutting force spectrum, and acoustic emission signals; based on Kalman filtering and inverse Fourier transform, it fuses multimodal data and vibration current signals to generate a comprehensive evaluation vector of the machining state; The adaptive control module comprises: Composite vibration control unit: outputs two phase-adjustable high-frequency sinusoidal signals through the ultrasonic power supply, drives the resonant ultrasonic knife through the addition circuit and power amplifier, and generates target trajectories such as sine waves and triangle waves; Dynamic parameter optimization unit: Combined with reinforcement learning algorithm, it adjusts vibration frequency, phase difference, cutting speed and feed rate in real time according to multi-modal feedback; Closed-loop feedback unit: Based on the current signal of the piezoelectric detection element, the resonant frequency is stabilized through frequency sweeping and impedance matching technology to suppress processing load fluctuations; The processing execution module integrates a multi-modal resonant ultrasonic vibration fast cutter, supports high-frequency axial intermittent motion, and realizes high-flexibility micro / nanostructure processing.

2. The multi-modal sensing-based precision parts adaptive processing system according to claim 1 is characterized in that: The composite vibration control unit includes an ultrasonic power supply, an adding circuit and a power amplifier, which are used to generate a composite excitation signal of a single frequency and a triple frequency to drive the resonant ultrasonic knife to generate multi-modal axial vibration.

3. The multi-modal sensing-based precision parts adaptive processing system according to claim 1 is characterized in that: The adaptive control module integrates a closed-loop feedback mechanism of a piezoelectric detection element and stabilizes the resonant frequency through frequency sweeping and impedance matching technology.

4. The multi-modal sensing-based precision parts adaptive processing system according to claim 1 is characterized in that: Inverse Fourier transform and reinforcement learning algorithm are used to achieve coordinated optimization of vibration trajectory and cutting parameters.

5. The method for adaptive processing of precision parts based on multimodal perception is characterized by: The following steps are involved: S1, synchronous acquisition of multimodal data: real-time acquisition of visual, force, acoustic, temperature data and vibration current signals; S2, high-frequency vibration trajectory generation: through the ultrasonic power supply to control the superposition of two frequency-doubled signals, the diamond tool is driven to generate the target waveform trajectory; S3, dynamic evaluation of machining status: integrating multi-modal features such as vibration phase difference, cutting force spectrum, surface roughness, etc. to determine tool wear level and thermal error; S4, closed-loop parameter optimization: According to the evaluation results, the amplitude-frequency characteristics of the vibration mode and the cutting parameters are adjusted, and the resonant frequency is calibrated in real time through piezoelectric detection feedback; S5, Iterative learning and process update: Optimize the deep learning model based on historical processing data to improve the generalization ability of adaptive control.

Images