Cooperative control method and equipment for thermal deformation of rotary chuck and medium

Abstract

The invention discloses a cooperative control method and device for thermal deformation of a rotary chuck and a medium, and relates to the technical field of semiconductors. The method comprises the following steps: collecting local temperature distribution and strain distribution data of the rotating base in an operation process; collecting a vibration response signal of the rotating base in the thermal-mechanical coupling state; frequency domain characteristic parameters are extracted from the vibration response signals; inputting the local temperature distribution, the strain distribution data and the frequency domain characteristic parameters into a pre-trained thermal deformation prediction model to obtain a thermal deformation sensitive area of the rotating base, a material performance degradation index and a thermal deformation compensation parameter; generating a corresponding pulse width modulation excitation current signal according to the thermal deformation compensation parameter; and applying an excitation current signal to the shape memory alloy grid of the thermal deformation sensitive area, and driving to generate compensation deformation opposite to the deformation direction. By means of the method, prediction of the thermal deformation trend and material performance degradation is achieved, and the control behavior is converted into active intervention from passive response.

Description

Technical Field

[0001] This application relates to the field of semiconductor technology, and in particular to a method, device and medium for the coordinated control of thermal deformation of a rotating chuck. Background Technology

[0002] In the semiconductor field, high-speed rotating bases are core components for achieving high-precision processes. During operation, uneven thermal deformation can occur due to factors such as friction and environmental temperature changes, which severely restricts processing accuracy and yield.

[0003] However, external cooling suffers from response lag and coarse temperature control; while uniformly distributed SMAs cannot accurately suppress dynamic, non-uniformly propagating thermal deformation, and lack the ability to perceive the evolution of material properties under long-term thermo-mechanical coupling, leading to the gradual failure of the compensation effect. The rich modal information contained in vibration signals is closely related to structural stress and material performance state, but this correlation has not yet been used to build a collaborative control system that can predict thermal deformation trends in real time and drive actuators for adaptive compensation.

[0004] Based on the above analysis, the problems and shortcomings of the existing technology are as follows: Existing technologies for high-speed rotating bases can only perceive and predict the material properties in real time, but cannot dynamically and accurately suppress the non-uniform thermal deformation of the high-speed rotating base. Summary of the Invention

[0005] This application provides a method, device, and medium for the coordinated control of thermal deformation of a rotating chuck. It can solve the problem that in the prior art, the high-speed rotating base cannot dynamically and accurately suppress the non-uniform thermal deformation of the high-speed rotating base because it cannot perceive and predict the material performance state in real time.

[0006] In a first aspect, embodiments of this application provide a collaborative control method for the thermal deformation of a rotating chuck. The method includes: acquiring local temperature and strain distribution data of the rotating base during operation using a multi-wavelength distributed fiber optic grating sensor array embedded inside the rotating base; acquiring vibration response signals of the rotating base under thermo-mechanical coupling using a piezoelectric ceramic sensor array arranged on the rotating base; extracting frequency domain feature parameters from the vibration response signals, including resonant frequency offset and modal damping ratio change; inputting the local temperature and strain distribution data and frequency domain feature parameters into a pre-trained thermal deformation prediction model to obtain the thermal deformation sensitive area of ​​the rotating base, as well as the material performance degradation index and thermal deformation compensation parameters; generating a corresponding pulse width modulation excitation current signal based on the thermal deformation compensation parameters; and applying the excitation current signal to a shape memory alloy mesh pre-embedded in the thermal deformation sensitive area of ​​the rotating base to drive compensation deformation opposite to the deformation direction.

[0007] In one implementation of this application, frequency domain feature parameters are extracted from the vibration response signal, specifically including: performing a fast Fourier transform on the acquired vibration response signal to obtain a frequency spectrum; identifying multiple characteristic peaks in the frequency spectrum corresponding to the base structure modes; calculating the offset of the characteristic peak frequency relative to the reference frequency as the resonant frequency offset; and calculating the resonant frequency offset and modal damping ratio change of each characteristic peak using the half-power bandwidth method.

[0008] In one implementation of this application, the thermal deformation prediction model is a hybrid model consisting of a convolutional neural network and a gated recurrent unit; the convolutional neural network is used to process spatially distributed temperature and strain data; and the gated recurrent unit is used to process the frequency domain feature parameters of the time series.

[0009] In one implementation of this application, a corresponding pulse width modulation excitation current signal is generated based on thermal deformation compensation parameters. Specifically, this includes: determining the amplitude, frequency, and phase of the current based on the thermal deformation compensation parameters; adaptively correcting the amplitude based on the material performance degradation index; and generating a pulse width modulation current waveform including amplitude, frequency, and phase based on a three-phase inverter circuit.

[0010] In one implementation of this application, the amplitude is adaptively corrected according to the material performance degradation index, specifically including: establishing a mapping relationship between the material performance degradation index and the driving efficiency of the shape memory alloy; when the material performance degradation index indicates that the material performance has degraded, increasing the amplitude of the excitation current according to a preset ratio to compensate for the decrease in driving efficiency.

[0011] In one implementation of this application, after driving the generation of compensating deformation opposite to the deformation direction, the method further includes: acquiring the compensated vibration response signal in real time through a piezoelectric ceramic sensor array; calculating the difference in characteristic parameters of the vibration response signal before and after compensation; if the difference does not reach the expected threshold, adjusting the input weights of the thermal deformation prediction model and regenerating the excitation current signal.

[0012] In one implementation of this application, before inputting the local temperature distribution, strain distribution data, and frequency domain feature parameters into the pre-trained thermal deformation prediction model, the method further includes: collecting historical data of the rotating base under various working conditions, the historical data including temperature distribution, strain distribution data, and vibration response signals; labeling the historical data, the labeled information including the degree of material performance degradation and ideal compensation parameters; and using the labeled historical data to perform supervised training on the hybrid model of convolutional neural network and gated recurrent unit.

[0013] In one implementation of this application, after driving the generation of compensating deformation opposite to the deformation direction, the method further includes: after the rotating base stops, applying a sweep frequency excitation signal through a piezoelectric ceramic sensor and acquiring a vibration response signal; analyzing the vibration response signal to evaluate the interface bonding state between the shape memory alloy mesh and the matrix material; and generating maintenance prompt information when the interface bonding state is lower than a set standard.

[0014] Secondly, embodiments of this application also provide a collaborative control device for the thermal deformation of a rotating chuck. The device includes at least one processor and a memory communicatively connected to the at least one processor. The memory stores instructions executable by the at least one processor, which, when executed, enable the at least one processor to: acquire local temperature and strain distribution data of the rotating base during operation using a multi-wavelength distributed fiber optic grating sensor array embedded within the rotating base; acquire vibration response signals of the rotating base under thermo-coupling conditions using a piezoelectric ceramic sensor array arranged on the rotating base; extract frequency domain feature parameters from the vibration response signals, including resonant frequency offset and modal damping ratio change; input the local temperature and strain distribution data and frequency domain feature parameters into a pre-trained thermal deformation prediction model to obtain the thermal deformation sensitive area of ​​the rotating base, as well as the material performance degradation index and thermal deformation compensation parameters; generate a corresponding pulse width modulation excitation current signal based on the thermal deformation compensation parameters; and apply the excitation current signal to a shape memory alloy mesh embedded in the thermal deformation sensitive area of ​​the rotating base to drive a compensation deformation opposite to the deformation direction.

[0015] Thirdly, this application also provides a non-volatile computer storage medium for the coordinated control of thermal deformation of a rotating chuck, storing computer-executable instructions. These instructions are configured to: acquire local temperature and strain distribution data of the rotating base during operation using a multi-wavelength distributed fiber optic grating sensor array embedded within the rotating base; acquire vibration response signals of the rotating base under thermo-mechanical coupling using a piezoelectric ceramic sensor array arranged on the rotating base; extract frequency domain feature parameters from the vibration response signals, including resonant frequency offset and modal damping ratio change; input the local temperature and strain distribution data and frequency domain feature parameters into a pre-trained thermal deformation prediction model to obtain a material performance degradation index and thermal deformation compensation parameters; generate a corresponding pulse width modulation excitation current signal based on the thermal deformation compensation parameters; and apply the excitation current signal to a shape memory alloy mesh pre-embedded in the thermal deformation sensitive area of ​​the rotating base to drive compensation deformation opposite to the deformation direction.

[0016] This application provides a collaborative control method, device, and medium for the thermal deformation of a rotating chuck. By fusing vibration signal analysis with a time-series deep learning model, it achieves prediction of thermal deformation trends and material performance degradation, transforming control behavior from passive response to active intervention. By extracting vibration frequency domain features and coupling them with the temperature-strain field, it constructs a non-contact, highly sensitive material state sensing capability, avoiding the integration of complex fluid structures inside the base and improving system reliability. Based on the prediction results, it dynamically generates and adaptively adjusts the excitation current applied to the SMA mesh, significantly improving the accuracy and adaptability of thermal deformation suppression, while extending the service life and maintenance cycle of the equipment. Attached Figure Description

[0017] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings: Figure 1 A flowchart illustrating a collaborative control method for the thermal deformation of a rotary chuck, provided in an embodiment of this application; Figure 2 This is a schematic diagram of the internal structure of a collaborative control device for thermal deformation of a rotary chuck, provided as an embodiment of this application. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0019] This application provides a method, device, and medium for the coordinated control of thermal deformation of a rotating chuck, which solves the problem in the prior art that the high-speed rotating base cannot dynamically and accurately suppress the non-uniform thermal deformation of the high-speed rotating base because it cannot perceive and predict the material performance state in real time.

[0020] The technical solutions proposed in the embodiments of this application will be described in detail below with reference to the accompanying drawings.

[0021] Figure 1 A flowchart illustrating a collaborative control method for the thermal deformation of a rotary chuck, provided in an embodiment of this application. Figure 1 As shown in the embodiment of this application, a collaborative control method for the thermal deformation of a rotary chuck specifically includes the following steps: Step 10: Collect local temperature and strain distribution data of the rotating base during operation using a multi-wavelength distributed fiber optic grating sensor array embedded inside the rotating base.

[0022] In this step, the temperature sensitivity of fiber Bragg gratings is utilized to collect local temperature distribution data in different areas of the base, reflecting the spatial differences in the temperature field under thermo-mechanical coupling. On the other hand, based on the strain sensing capability of fiber Bragg gratings, strain distribution information of various parts of the base is obtained, characterizing the deformation state of the structure caused by the combined effects of temperature changes and mechanical loads. Depending on the structural dimensions of the rotating base and the coverage requirements of the thermal deformation sensitive area, 6-12 multi-wavelength distributed fiber Bragg grating sensors can be configured, with each fiber containing 3-5 grating measurement points, for a total of 18-60 measurement points, ensuring... No blind spots in monitoring; 4-6 optical fibers are evenly arranged along the circumference of the chuck jaw mounting base, with each fiber layered radially to cover the transition area between the jaw base and the base body, capturing the high strain and high temperature gradient region caused by the combined effects of clamping force and frictional heat; 2 optical fibers are embedded radially along the mating surface between the base and the spindle to monitor the axial temperature and strain distribution caused by spindle heat transfer; 1-2 optical fibers are arranged in the non-sensitive area of ​​the base as reference measuring points to offset the influence of environmental interference on core data. Drilling and pre-embedding and high-temperature epoxy curing can be used to ensure that the optical fibers are tightly bonded to the base material without relative displacement.

[0023] Step 20: Collect the vibration response signal of the rotating base under thermo-coupling state by using a piezoelectric ceramic sensor array arranged on the rotating base.

[0024] In this step, the piezoelectric ceramic sensor operates based on the piezoelectric effect. A flexible piezoelectric sensor can be considered, which can convert the mechanical vibration displacement, velocity, or acceleration of the base into a measurable electrical signal. It has the characteristics of fast response speed and wide dynamic range. The sensor is arranged in an array on the surface, edge, or integrated structure, depending on the actual situation. It can cover the key stress areas and easily deformable parts of the base, realize multi-point synchronous acquisition, avoid the limitations of single-point monitoring, and thus fully capture the overall vibration characteristics of the base under thermo-mechanical coupling, that is, under the combined action of temperature field and mechanical load.

[0025] The thermo-coupling state here refers to the combined effect of temperature changes, including frictional heat and ambient temperature difference, on the base during operation, along with the centrifugal force of mechanical load rotation and the clamping force of the workpiece. The vibration response signal under this state contains information on the changes in dynamic parameters such as structural stiffness and damping, providing a basis for subsequent extraction of frequency domain features.

[0026] Understandably, after data collection, it is necessary to perform noise reduction and repair, and preprocessing such as removing anomalies and filling in missing data is required.

[0027] Furthermore, during the operation of the rotary chuck, especially in working environments with corrosive media such as etching processes, the rotary chuck needs to have good corrosion resistance to ensure the long-term reliable operation of its key components in acid and alkali environments. In this case, the sensor can be treated with corrosion resistance, such as using polytetrafluoroethylene coating or sealing, and using nickel-titanium alloy for the shape memory alloy mesh, which is itself resistant to corrosion by neutral and weak acid solutions.

[0028] Step 30: Extract frequency domain characteristic parameters from the vibration response signal. The frequency domain characteristic parameters include the resonant frequency offset and the change in modal damping ratio.

[0029] In this step, the thermal deformation of the structure will cause changes in dynamic characteristics such as stiffness and damping, and these changes will be directly reflected in the frequency characteristics of the vibration response. By extracting the resonant frequency shift and the change in modal damping ratio, this characteristic change can be quantitatively characterized.

[0030] As an optional embodiment, extracting frequency domain feature parameters from the vibration response signal may specifically include: Step 301: Perform a Fast Fourier Transform on the acquired vibration response signal to obtain the frequency spectrum. In this step, the vibration response signal acquired by the piezoelectric ceramic sensor, which is the time-domain signal, i.e. the vibration amplitude that changes with time, is subjected to a Fast Fourier Transform to convert the signal from the time-amplitude domain to the frequency-energy domain, thereby obtaining the frequency spectrum. This spectrum can intuitively present the frequency components contained in the signal and their corresponding energy intensities.

[0031] Step 302: Identify multiple characteristic peaks in the frequency spectrum that correspond to the modes of the base structure; In this step, multiple characteristic peaks corresponding to the structural modes of the rotating base are identified in the obtained frequency spectrum. The structural modes refer to the inherent vibration modes of the base as an elastic body, such as first-order bending, second-order torsion, etc. Each mode corresponds to a natural frequency, which is represented by a characteristic peak with concentrated energy in the frequency spectrum. The higher the peak value, the stronger the vibration energy at that frequency. By identifying these characteristic peaks, the frequency position of the key vibration modes of the base can be located.

[0032] Step 303: Calculate the offset of the characteristic peak frequency relative to the reference frequency, as the resonant frequency offset; In this step, the natural frequency of the base in its normal state, that is, under conditions of no significant thermal deformation and stable material properties, is taken as the reference frequency. The difference between the characteristic peak frequency identified in step 302 and the corresponding reference frequency is calculated as the resonant frequency offset. This offset directly reflects the change in structural stiffness caused by thermal deformation. Thermal deformation will change the geometric dimensions or material properties of the structure, thereby causing the natural frequency to shift. For example, when the stiffness decreases, the resonant frequency usually shifts to the lower frequency direction.

[0033] Step 304: Calculate the resonant frequency shift and modal damping ratio change of each characteristic peak using the half-power bandwidth method.

[0034] In this step, the half-power bandwidth method is used to calculate the modal damping ratio corresponding to each characteristic peak: On the frequency spectrum, the two frequency values ​​corresponding to 0.707 times the peak value of the characteristic peak (half-power point) are taken, and the difference between the two is the half-power bandwidth. The ratio of this bandwidth to the center frequency of the characteristic peak is the modal damping ratio. Subsequently, the calculated damping ratio is compared with the damping ratio under the reference state to obtain the change in modal damping ratio. This change reflects the change in the structure's energy dissipation capacity. Changes in the local contact state of the structure caused by thermal deformation may increase or decrease the damping ratio.

[0035] Step 40: Input the local temperature distribution, strain distribution data and frequency domain feature parameters into the pre-trained thermal deformation prediction model to obtain the thermal deformation sensitive area of ​​the rotating base, as well as the material property degradation index and thermal deformation compensation parameters; wherein, the thermal deformation prediction model is a hybrid model composed of a convolutional neural network and a gated recurrent unit; the convolutional neural network is used to process the spatially distributed temperature and strain data; the gated recurrent unit is used to process the frequency domain feature parameters of the time series.

[0036] In this step, the gate processes the spatially distributed temperature and strain data. The convolutional neural network, through structures such as convolutional layers and pooling layers, can automatically extract local spatial correlation features in the data, such as temperature gradient distribution and strain concentration areas, and capture the physical field coupling relationship of different parts of the base in space: the feature of a sudden temperature rise in a certain area accompanied by significant strain concentration, thereby transforming high-dimensional spatial distribution data into low-dimensional spatial feature vectors with physical meaning.

[0037] Gated recurrent units are specifically designed to process the frequency domain feature parameters of time series. As an improved structure of recurrent neural networks, they effectively capture long-term dependencies in time series through gating mechanisms. They can analyze the dynamic evolution of frequency domain feature parameters over time, the continuous shift trend of resonant frequency with running time, and the fluctuation pattern of modal damping ratio, transforming time series data into time feature vectors that reflect dynamic changes.

[0038] Polar coordinate grid data with the base center as the origin needs to be converted into a spatial matrix and normalized. Time-series data needs to be converted into a time-series matrix and standardized. Based on a hybrid model of convolutional neural networks and gated recurrent units, deep correlation of multi-source data is achieved through three levels of processing: spatial feature extraction, temporal feature extraction, and cross-modal fusion. Spatial feature extraction: Local temperature and strain distribution data are input into the CNN branch. The convolutional layer extracts local spatial correlation features through 3×3 convolutional kernels. The pooling layer retains key features and reduces dimensionality. The fully connected layer compresses the spatial features into a 128-dimensional spatial feature vector, including sensitive area location and temperature-strain coupling strength. Temporal feature extraction: Frequency domain feature parameters are input into the GRU branch. The GRU layer captures temporal dependencies through a gating mechanism, such as the continuous shift trend of resonant frequency with increasing temperature and the transient change law of modal damping ratio under sudden load changes. The fully connected layer compresses the temporal features into a 64-dimensional temporal feature vector, including information such as the degradation rate of structural dynamic characteristics and the mapping relationship between frequency domain parameters and thermal deformation. Cross-modal feature fusion integrates spatial and temporal feature vectors through a concatenation and attention mechanism: the concatenated feature vector is input into the attention layer, where the model automatically assigns weights, such as higher weights for spatial features in high-temperature, high-strain regions and higher weights for temporal features during periods of abrupt changes in frequency domain parameters; the fused feature vector is input into three parallel output heads, corresponding to three target outputs. The fused features generate specific results through the three output heads. Output head 1 for thermal deformation sensitive regions contains three fully connected layers and outputs an N×N binary matrix, where N is the side length of the base grid: 1 in the matrix indicates that the grid cell is a sensitive region, and 0 indicates a non-sensitive region; based on high temperature gradient and high strain features extracted by CNN, combined with frequency domain parameter anomaly regions captured by GRU, the spatial distribution of the output sensitive region is determined by a threshold.

[0039] Material property degradation index output head 2 contains two fully connected layers, with an output dimension of 1 and a numerical range of 0-1: an index of 0 indicates intact material properties, and 1 indicates severe degradation. Based on the long-term evolution trend of frequency domain parameters extracted by GRU, the modal damping ratio continuously increases with running time, combined with strain accumulation features extracted by CNN, such as material fatigue features caused by cyclic strain, and outputs a quantized index through the degradation-feature mapping relationship learned by the model. Thermal deformation compensation parameter output head 3 contains two fully connected layers, with an output dimension of 3, corresponding to current amplitude, frequency, and phase respectively. Amplitude: 0.5-2.0A, positively correlated with the thermal deformation of the sensitive area; Frequency: 50-200Hz, matching the thermal response rate of SMA mesh; Phase: 0-360°, controlling the coordinated action of multi-region SMA meshes; By integrating the spatial distribution of sensitive areas, determining the compensation range, the material performance degradation index, and correcting the amplitude (the more severe the degradation, the larger the amplitude), and considering the dynamic changes in frequency domain characteristics, the frequency and phase are adjusted to adapt to the real-time deformation rate, and the output is the optimal parameter that conforms to the compensation deformation amount = SMA recovery amount.

[0040] Step 50: Generate the corresponding pulse width modulation excitation current signal based on the thermal deformation compensation parameters; As an optional embodiment, a corresponding pulse width modulation excitation current signal is generated based on the thermal deformation compensation parameters, which may specifically include: Step 501: Determine the amplitude, frequency and phase of the current based on the thermal deformation compensation parameters; In this step, based on the thermal deformation compensation parameters, which reflect the magnitude, rate, and direction of the required compensation deformation, three core parameters of the pulse width modulation excitation current are determined: Amplitude: corresponds to the current intensity, which directly determines the heating power and deformation of the shape memory alloy. The larger the amplitude, the faster the heating and the greater the deformation; Frequency: corresponds to the alternation rate of the current, which needs to match the thermal response characteristics of the shape memory alloy, such as the phase transformation rate of the alloy, to ensure that the deformation process is stable and controllable; Phase: is used for the timing synchronization when multiple sets of shape memory alloy meshes work together. Alloys in different regions need to be deformed in a specific order to avoid interference.

[0041] Step 502: Adaptively correct the amplitude based on the material property degradation index; In this step, the current amplitude determined in step 501 is dynamically adjusted by combining the material performance degradation index to reflect the degree of performance degradation of the shape memory alloy or base material. Material performance degradation: the decrease in alloy phase transformation efficiency and the decrease in elastic modulus will lead to a decrease in the actual deformation under the same current. Therefore, it is necessary to establish a mapping relationship between the degradation index and the driving efficiency. When the degradation index exceeds the set threshold, the current amplitude is increased by a preset ratio, which can be a linear ratio or a piecewise ratio, to compensate for the efficiency loss and ensure that the actual deformation is consistent with the theoretical requirement.

[0042] Step 503: Generate a pulse width modulated current waveform including amplitude, frequency and phase based on the three-phase inverter circuit.

[0043] In this step, based on the three-phase inverter circuit, the above-mentioned corrected amplitude, frequency, and phase parameters are converted into actual pulse width modulation current waveforms. The three-phase inverter circuit converts DC power into AC power that meets the parameter requirements through the high-frequency switching of power switching devices. The pulse width modulation technology achieves precise control of the average current amplitude by controlling the duty cycle of the switching. The switching frequency is much higher than the thermal response frequency of the alloy, which can be approximated as a continuous and stable current output. This method can ensure the accuracy of current parameters and improve energy conversion efficiency, making it suitable for dynamic power supply scenarios of rotating bases.

[0044] As an optional embodiment, the amplitude is adaptively corrected according to the material performance degradation index, which may specifically include: step 5021: establishing a mapping relationship between the material performance degradation index and the driving efficiency of the shape memory alloy; step 5022: when the material performance degradation index indicates that the material performance has degraded, increasing the amplitude of the excitation current according to a preset ratio to compensate for the decrease in driving efficiency.

[0045] Step 60: Apply an excitation current signal to the shape memory alloy mesh in the heat deformation sensitive area to drive a compensating deformation opposite to the deformation direction.

[0046] In this step, when the excitation current signal is applied to the SMA (Shape Memory Alloy), the current generates Joule heat through the alloy material, causing the mesh temperature to rise and triggering a phase transition. Since the preset shape of the SMA mesh is designed based on the thermal deformation trend of the base, if the base tends to bend in a certain direction after being heated, including radial expansion due to thermal effects, edge expansion due to centrifugal force, local warping due to temperature gradient, etc., the memory shape of the SMA mesh is preset to bend in the opposite direction. The recovery deformation generated during the phase transition will directly act on the sensitive area of ​​the base, forming a mechanical force opposite to the direction of thermal deformation, thereby offsetting or weakening the original thermal deformation of the base.

[0047] Furthermore, Ni55Ti45 nickel-titanium shape memory alloy can be selected, as its phase transition temperature is well-suited to match the operating temperature range of the base, avoiding low-temperature incompatibility or high-temperature over-deformation. The grid units are woven from SMA wires; 2-3 layers of orthogonal woven grids can be used, fitting snugly against the base surface and corresponding one-to-one with the core sensitive areas of the fiber optic sensor, namely the radial area below the chuck jaw mounting base and the spindle connection end. Each sensitive area is covered by 1-2 grids to avoid gaps; areas with thermal deformation greater than 0.3mm can also use 3 layers of grids to enhance compensation force. The grid is pre-embedded 0.5-1mm below the base surface, through die casting or subsequent slotting embedding, with the upper surface of the grid flush with the base surface to avoid affecting the chuck jaw installation; the grid edges are connected to the base metal substrate via SMA solder joints or encapsulated with high-temperature resin to ensure synchronous deformation of the grid and base during phase transition, without relative slippage; fine copper electrodes are welded to both ends of the grid, connected to slip rings through internal base wires to achieve stable transmission of excitation current.

[0048] Furthermore, the distributed arrangement of the SMA mesh enables area compensation for thermal deformation, avoiding stress concentration caused by single-point compensation. At the same time, its deformation response speed matches the control precision of the current signal. Through the pulse width modulation parameters in step 50, dynamic and real-time compensation can be achieved, ultimately maintaining the structural accuracy of the rotating base under thermo-mechanical coupling conditions.

[0049] As an optional embodiment, after driving the generation of compensating deformation opposite to the deformation direction, the method may further include: acquiring the compensated vibration response signal in real time through a piezoelectric ceramic sensor array; calculating the difference in characteristic parameters of the vibration response signal before and after compensation; if the difference does not reach the expected threshold, adjusting the input weights of the thermal deformation prediction model and regenerating the excitation current signal.

[0050] In this step, the feature extraction method from step 30 is used to extract the resonant frequency offset and modal damping ratio change from the acquired post-compensation vibration response signal. Then, the difference between these two feature parameters before and after compensation is calculated. This difference is used to quantify and evaluate the improvement effect of the compensation measures on the dynamic characteristics of the base. If the compensation is effective, the resonant frequency offset should revert to the reference frequency, and the modal damping ratio change should approach a stable range. The calculated difference in feature parameters is compared with a preset expected threshold, pre-set according to the accuracy requirements and operating conditions of the rotating chuck, such as the resonant frequency offset returning to the reference value. If the difference does not reach the expected threshold, it indicates that the current compensation effect is not satisfactory. The demand may be due to unreasonable weighting of multi-source data in the model, which may lead to deviations in the output compensation parameters. In this case, it is necessary to dynamically adjust the input weights of the thermal deformation prediction model and optimize the contribution ratio of the convolutional neural network and the gated recurrent unit to the model output. For example, if the temperature distribution data has a more significant impact on the compensation effect, the input weights should be increased. The adjusted thermal deformation prediction model re-receives the current local temperature distribution, strain distribution data and the compensated frequency domain feature parameters, and outputs updated thermal deformation compensation parameters. Then, the excitation current signal is regenerated according to step 50, and the shape memory alloy mesh is driven to perform a new round of compensation according to step 60 until the difference in feature parameters reaches the expected threshold.

[0051] As an optional embodiment, before inputting the local temperature distribution, strain distribution data, and frequency domain feature parameters into the pre-trained thermal deformation prediction model, the method may further include: Step 01: Collecting historical data of the rotating base under various working conditions, the historical data including temperature distribution, strain distribution data, and vibration response signals; Step 02: Labeling the historical data, the labeled information including the degree of material performance degradation and ideal compensation parameters; Step 03: Using the labeled historical data to perform supervised training on the hybrid model of convolutional neural network and gated recurrent unit.

[0052] In this step, during the commissioning or trial operation of the rotating base, various typical working conditions that it may encounter in actual operation are simulated: different rotation speeds, load intensities, ambient temperatures, running times, etc. Historical data under the corresponding working conditions are collected synchronously through fiber optic grating sensor arrays and piezoelectric ceramic sensor arrays. By collecting data under diverse working conditions, the model can learn the laws of thermal deformation under different scenarios, avoiding insufficient generalization ability due to single data.

[0053] The collected historical data is manually or automatically labeled to supplement two types of key label information: the degree of material performance degradation: the performance degradation level of the rotating base material under the corresponding working condition is determined by material testing methods such as hardness testing, microstructure analysis or indirect characterization methods; ideal compensation parameters: the optimal compensation parameters that can completely offset thermal deformation under the working condition are determined by simulation calculation or actual debugging and verification, including the current amplitude, frequency, phase and other parameters required for the shape memory alloy mesh.

[0054] The essence of annotation is to assign an input-output mapping relationship to historical data. Using the annotated historical data as training samples, supervised training is performed on a hybrid model composed of a convolutional neural network and a gated recurrent unit (GRU): temperature distribution and strain distribution data are input into the CNN to learn the correlation between spatial features and material degradation and compensation parameters; the time-series data of frequency domain feature parameters converted from vibration response signals are input into the GRU to learn the time-series evolution law of dynamic features and the mapping to the output target; the outputs of the two types of sub-models are integrated through a feature fusion layer, and the predicted material performance degradation index and thermal deformation compensation parameters are output through a fully connected layer; the error between the predicted value and the annotated value (mean square error) is used as the loss function, and the model parameters, namely the convolutional kernel weights of the CNN and the gating parameters of the GRU, are iteratively optimized through the backpropagation algorithm until the loss function converges to a preset threshold, thus completing the model training.

[0055] As an optional embodiment, after driving the generation of compensating deformation opposite to the deformation direction, the method may further include: after the rotating base stops, applying a sweep frequency excitation signal through a piezoelectric ceramic sensor and acquiring a vibration response signal; analyzing the vibration response signal to evaluate the interface bonding state between the shape memory alloy mesh and the matrix material; and generating maintenance prompt information when the interface bonding state is lower than a set standard.

[0056] In this step, some piezoelectric ceramic sensors are selected as exciters to apply a frequency sweep excitation signal within a preset frequency range to the base. The frequency range needs to cover the key natural frequencies of the base and the SMA mesh working together, typically scanning continuously from low to high frequencies. The remaining piezoelectric ceramic sensors act as receivers, synchronously acquiring the vibration response signal of the base under the frequency sweep excitation. The natural frequencies, modal damping ratios, and mode shapes of the base in the current state are extracted using a modal recognition algorithm and compared with the reference state. If there is debonding / loosening at the interface, it will cause a decrease in the overall stiffness of the structure, manifested as a shift of the natural frequencies towards lower frequencies. At the same time, interface friction or gaps will increase energy dissipation, resulting in a significant increase in the modal damping ratio; and local mode shape abrupt changes may occur.

[0057] Calculate the transfer function between the excitation point and the response point, and output the ratio of the response to the input excitation. If the interface is poorly bonded, the peak value of the transfer function will become wider and the amplitude will decrease, the vibration energy transfer will be hindered, and even additional stray peaks will appear, resulting in local vibration at the interface.

[0058] The degree of anomaly of the above characteristic parameters is used to quantitatively evaluate the interface bonding state between the SMA mesh and the substrate. The evaluated interface bonding state is then compared with the preset maintenance standard: if the interface bonding state meets the standard, it is determined that the subsequent compensation function can be implemented normally and no maintenance is required; if the interface bonding state is below the standard, there is serious debonding, and the system automatically generates targeted maintenance prompts.

[0059] The above are embodiments of the method proposed in this application. Based on the same inventive concept, embodiments of this application also provide a collaborative control device for the thermal deformation of a rotary chuck, the structure of which is as follows: Figure 2 As shown.

[0060] Figure 2 This is a schematic diagram of the internal structure of a collaborative control device for the thermal deformation of a rotary chuck, provided as an embodiment of this application. Figure 2 As shown, the device includes: At least one processor 201; And a memory 202 that is communicatively connected to at least one processor; The memory 202 stores instructions executable by at least one processor 201, which in turn executes the instructions to enable the processor 201 to: acquire local temperature and strain distribution data of the rotating base during operation using a multi-wavelength distributed fiber optic grating sensor array embedded within the rotating base; acquire vibration response signals of the rotating base under thermo-coupling conditions using a piezoelectric ceramic sensor array arranged on the rotating base; extract frequency domain feature parameters from the vibration response signals, including resonant frequency offset and modal damping ratio change; input the local temperature and strain distribution data and frequency domain feature parameters into a pre-trained thermal deformation prediction model to obtain the thermal deformation sensitive area of ​​the rotating base, as well as the material performance degradation index and thermal deformation compensation parameters; generate a corresponding pulse width modulation excitation current signal based on the thermal deformation compensation parameters; and apply the excitation current signal to a shape memory alloy mesh embedded in the thermal deformation sensitive area of ​​the rotating base to drive a compensation deformation opposite to the deformation direction.

[0061] Some embodiments of this application provide corresponding to Figure 1A non-volatile computer storage medium for the coordinated control of thermal deformation of a rotating chuck stores computer-executable instructions. These instructions are configured to: acquire local temperature and strain distribution data of the rotating base during operation using a multi-wavelength distributed fiber optic grating sensor array embedded within the rotating base; acquire vibration response signals of the rotating base under thermo-mechanical coupling using a piezoelectric ceramic sensor array arranged on the rotating base; extract frequency domain feature parameters from the vibration response signals, including resonant frequency offset and modal damping ratio change; input the local temperature and strain distribution data and frequency domain feature parameters into a pre-trained thermal deformation prediction model to obtain the thermal deformation sensitive area of ​​the rotating base, as well as the material performance degradation index and thermal deformation compensation parameters; generate a corresponding pulse width modulation excitation current signal based on the thermal deformation compensation parameters; and apply the excitation current signal to a shape memory alloy mesh pre-embedded in the thermal deformation sensitive area of ​​the rotating base to drive compensation deformation opposite to the deformation direction.

[0062] The various embodiments in this application are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the embodiments for IoT devices and media are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0063] The systems, media, and methods provided in this application are one-to-one correspondences. Therefore, the systems and media also have similar beneficial technical effects as their corresponding methods. Since the beneficial technical effects of the methods have been described in detail above, the beneficial technical effects of the systems and media will not be repeated here.

[0064] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0065] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0066] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0067] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0068] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0069] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0070] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0071] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0072] The above description is merely an embodiment of this application and is not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A method for coordinated control of thermal deformation of a rotary chuck, characterized in that, The method includes: The local temperature and strain distribution data of the rotating base during operation are collected by a multi-wavelength distributed fiber optic grating sensor array embedded inside the rotating base. The vibration response signal of the rotating base under thermo-mechanical coupling state is collected by an array of piezoelectric ceramic sensors arranged on the rotating base. Frequency domain feature parameters are extracted from the vibration response signal, including resonant frequency offset and modal damping ratio change. The local temperature distribution, strain distribution data and frequency domain feature parameters are input into a pre-trained thermal deformation prediction model to obtain the thermal deformation sensitive area of ​​the rotating base, as well as the material property degradation index and thermal deformation compensation parameters. Based on the thermal deformation compensation parameters, a corresponding pulse width modulation excitation current signal is generated; The excitation current signal is applied to the shape memory alloy mesh in the heat deformation sensitive area to drive the generation of compensating deformation opposite to the deformation direction.

2. The method for coordinated control of thermal deformation of a rotary chuck according to claim 1, characterized in that, Extracting frequency domain feature parameters from the vibration response signal specifically includes: The collected vibration response signal is subjected to a fast Fourier transform to obtain the frequency spectrum; Identify multiple characteristic peaks in the frequency spectrum that correspond to the modes of the base structure; Calculate the offset of the characteristic peak frequency relative to the reference frequency, and use it as the resonant frequency offset; The resonant frequency shift and modal damping ratio change of each characteristic peak were calculated using the half-power bandwidth method.

3. The method for coordinated control of thermal deformation of a rotary chuck according to claim 1, characterized in that, The thermal deformation prediction model is a hybrid model composed of a convolutional neural network and a gated recurrent unit; The convolutional neural network is used to process spatially distributed temperature and strain data; The gated loop unit is used to process the frequency domain feature parameters of the time series.

4. The method for coordinated control of thermal deformation of a rotary chuck according to claim 1, characterized in that, The step of generating a corresponding pulse width modulation excitation current signal based on the thermal deformation compensation parameters specifically includes: The amplitude, frequency, and phase of the current are determined based on the thermal deformation compensation parameters. The amplitude is adaptively corrected based on the material property degradation index; A pulse-width modulated current waveform, including the amplitude, frequency, and phase, is generated based on a three-phase inverter circuit.

5. The method for coordinated control of thermal deformation of a rotary chuck according to claim 4, characterized in that, The adaptive correction of the amplitude based on the material property degradation index specifically includes: Establish a mapping relationship between the material property degradation index and the shape memory alloy driving efficiency; When the material performance degradation index indicates that the material performance has degraded, the amplitude of the excitation current is increased according to a preset ratio to compensate for the decrease in driving efficiency.

6. The method for coordinated control of thermal deformation of a rotary chuck according to claim 1, characterized in that, After the drive generates a compensating deformation opposite to the deformation direction, the method further includes: The compensated vibration response signal is acquired in real time using the piezoelectric ceramic sensor array. Calculate the differences in characteristic parameters of the vibration response signal before and after compensation; If the difference does not reach the expected threshold, the input weights of the thermal deformation prediction model are adjusted, and the excitation current signal is regenerated.

7. The method for coordinated control of thermal deformation of a rotary chuck according to claim 1, characterized in that, Before inputting the local temperature distribution, strain distribution data, and frequency domain feature parameters into the pre-trained thermal deformation prediction model, the method further includes: Historical data of the rotating base were collected under various working conditions, including temperature distribution, strain distribution data, and vibration response signals. The historical data is annotated, and the annotated information includes the degree of material performance degradation and ideal compensation parameters; The hybrid model of the convolutional neural network and gated recurrent unit was trained in a supervised manner using labeled historical data.

8. The method for coordinated control of thermal deformation of a rotary chuck according to claim 1, characterized in that, After the drive generates a compensating deformation opposite to the deformation direction, the method further includes: After the rotating base stops, a sweep frequency excitation signal is applied through a piezoelectric ceramic sensor, and the vibration response signal is collected. Analyze the vibration response signal to evaluate the interfacial bonding state between the shape memory alloy mesh and the matrix material; When the interface connection status falls below the set standard, a maintenance prompt message is generated.

9. A device, characterized in that, The device includes: At least one processor; And, a memory communicatively connected to the at least one processor; The memory stores instructions executable by the at least one processor, which, when executed by the at least one processor, enable the at least one processor to: The local temperature and strain distribution data of the rotating base during operation are collected by a multi-wavelength distributed fiber optic grating sensor array embedded inside the rotating base. The vibration response signal of the rotating base under thermo-mechanical coupling state is collected by an array of piezoelectric ceramic sensors arranged on the rotating base. Frequency domain feature parameters are extracted from the vibration response signal, including resonant frequency offset and modal damping ratio change. The local temperature distribution, strain distribution data and frequency domain feature parameters are input into a pre-trained thermal deformation prediction model to obtain the thermal deformation sensitive area of ​​the rotating base, as well as the material property degradation index and thermal deformation compensation parameters. Based on the thermal deformation compensation parameters, a corresponding pulse width modulation excitation current signal is generated; The excitation current signal is applied to the shape memory alloy mesh in the heat deformation sensitive area to drive the generation of compensating deformation opposite to the deformation direction.

10. A non-volatile computer storage medium storing computer-executable instructions, characterized in that, The computer-executable instructions are set as follows: The local temperature and strain distribution data of the rotating base during operation are collected by a multi-wavelength distributed fiber optic grating sensor array embedded inside the rotating base. The vibration response signal of the rotating base under thermo-mechanical coupling state is collected by an array of piezoelectric ceramic sensors arranged on the rotating base. Frequency domain feature parameters are extracted from the vibration response signal, including resonant frequency offset and modal damping ratio change. The local temperature distribution, strain distribution data and frequency domain feature parameters are input into a pre-trained thermal deformation prediction model to obtain the thermal deformation sensitive area of ​​the rotating base, as well as the material property degradation index and thermal deformation compensation parameters. Based on the thermal deformation compensation parameters, a corresponding pulse width modulation excitation current signal is generated; The excitation current signal is applied to a shape memory alloy mesh embedded in the heat deformation sensitive area of ​​the rotating base to drive a compensating deformation opposite to the deformation direction.

Images