Intelligent servo control method for vertical superfinishing machine based on dynamic parameter identification

Abstract

The application discloses a vertical super-precision machine intelligent servo control method based on dynamic parameter identification and relates to the technical field of servo control. The method comprises the following steps: arranging a first sensing device on a key motion shaft of a target device, so as to acquire vibration data of the motion shaft and obtain vibration information items; and arranging a second sensing device on an oil stone unit, wherein the second sensing device is a pressure and temperature integrated sensing device, so as to acquire real-time pressure data and temperature data of the oil stone and obtain pressure information items and temperature information items. According to the application, low-damage combinations of different oscillation frequencies and feed speeds are preferentially selected as the device to be executed through combined damage sequencing, and the update information of the vibration information items, the pressure information items and the temperature information items is acquired in a deep level, so that the initial information items are secondarily adjusted, thereby realizing the oil stone driving control based on the dynamic parameter identification of the combination of the vibration data, the temperature data and the pressure data.

Description

Technical Field

[0001] This invention relates to the field of servo control technology, specifically to an intelligent servo control method for vertical ultraprecision machines based on dynamic parameter identification. Background Technology

[0002] Vertical ultra-precision machine tools refer to ultra-precision machine tools with a vertical structure, often used for high-precision machining tasks such as grinding, polishing, and micro-engraving. Servo control is a feedback-based control system that uses servo motors and sensors to adjust the output in real time, ensuring precise movements and rapid response. The common basic method is PID control. The intelligent servo control method for vertical ultra-precision machine tools refers to an advanced control system applied to vertical ultra-precision machine tools. It combines servo control technology and artificial intelligence algorithms to achieve higher precision, efficiency, and adaptive machining processes. It is a technical method that optimizes the operation of servo motors through intelligent algorithms, thereby improving the performance and reliability of ultra-precision machine tools.

[0003] A variable gain control method for the servo loop of a three-axis inertial stabilization platform system, disclosed in patent publication number CN111624873A, overcomes the problem of decreased servo loop stability caused by changes in system bandwidth due to variations in the moment of inertia when βyk and βzk are non-zero under all attitude conditions. This method achieves robust stability under all attitude conditions. Furthermore, it overcomes the problem of decreased static and dynamic accuracy of the servo loop caused by changes in the moment of inertia when βyk and βzk are non-zero under all attitude conditions, achieving accuracy retention under all attitude conditions. Moreover, it eliminates unsolvable regions in the calculation process, covering arbitrary attitude angles, and is more accurate and widely applicable than existing calculation methods.

[0004] In the process of servo control, the above-mentioned and similar technical solutions require high precision when machining workpieces using vertical ultra-precision machines. However, due to long-term mechanical wear, the internal components such as guide rails, ball screws, and oilstones will experience precision drift due to wear, which further affects the precision. Furthermore, the oscillation data caused by the oilstone during machining varies depending on the oscillation frequency and feed speed. Therefore, under the combined influence of a fixed oscillation frequency and feed speed, it is impossible to obtain the minimum precision drift. Summary of the Invention

[0005] The purpose of this invention is to provide an intelligent servo control method for vertical ultra-precision machines based on dynamic parameter identification, so as to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a method for intelligent servo control of a vertical ultraprecision machine based on dynamic parameter identification, comprising:

[0007] A first sensing device is arranged on the key motion axis of the target device to acquire vibration data of the motion axis and obtain vibration information items.

[0008] A second sensing device is installed on the oilstone unit. The second sensing device is an integrated pressure and temperature sensing device, which is used to acquire real-time pressure data and temperature data of the oilstone to obtain pressure information items and temperature information items.

[0009] The processing requirements are obtained, which are the accuracy requirements of the processing surface. Based on the processing requirements, the actual accuracy data of the target workpiece is obtained through image acquisition equipment to obtain the real-time accuracy item.

[0010] Acquire initial data of the oilstone, including oscillation frequency and feed rate, to obtain initial information items. Based on the difference between the real-time accuracy item and the processing requirements, create a first correlation model based on the initial information. The first correlation model outputs correlation data between different initial data and processing surface accuracy data. Based on the processing requirements as input, perform preliminary adjustment on the initial information items through the first correlation model to obtain initial adjustment items.

[0011] Based on the initial adjustment term, the updated information of vibration, pressure and temperature information terms is obtained. The initial information terms are then adjusted a second time using the adjustment method, thereby realizing the oilstone drive control based on dynamic parameter identification of the combination of vibration data, temperature data and pressure data.

[0012] Furthermore, the first sensing device is a vibration sensing device, and the method for acquiring vibration information items includes:

[0013] Obtain information on the motion axis components of the target vertical ultraprecision machine, including the oilstone supply guide rail, lead screw, and C-axis, and obtain the installation information items;

[0014] Based on the installation information items, an interval value is set. The interval value is a fixed value. The installation information items are divided into regions based on the interval value to obtain at least three region division items. Vibration sensing devices are arranged according to the region division items to obtain vibration data and obtain a vibration dataset.

[0015] The average vibration information of the vibration dataset is obtained, and then the vibration information item is obtained.

[0016] Furthermore, the second sensing device is a flexible multimodal sensing patch, and the methods for acquiring pressure and temperature information items include:

[0017] Acquire oilstone working scene information, including workpiece contact area and oscillation area, and divide the oilstone working scene information into regions based on workpiece contact area and oscillation area to obtain first region and second region;

[0018] Based on the first region and the second region, flexible sensing patches are attached respectively. The flexible sensing patches integrate a microcavity pressure sensor and a thin-film thermocouple sensor to obtain installation information items.

[0019] Based on the installation information items, pressure data and temperature data are obtained, resulting in pressure information items and temperature information items.

[0020] Furthermore, the method for obtaining the processing requirements includes:

[0021] Acquire the target workpiece machining data, including machining drawings; perform information scanning based on the machining data to obtain precision annotation data and obtain annotation information items;

[0022] Based on the labeled information items, processing parameter instructions are generated. These processing parameter instructions are then used as input to the target vertical ultraprecision machine to obtain output data, thereby obtaining the processing requirements.

[0023] Furthermore, the image acquisition device includes a non-contact detection probe, and the method for obtaining the real-time accuracy term includes:

[0024] The non-contact detection probe includes a confocal microscopic probe assembly with red-blue-ultraviolet three-bands, which is used to penetrate oil mist to obtain nanoscale three-dimensional morphology and acquire processing images. Based on the processing images, placement points are set, which are integrated on the side of the oilstone. The non-contact probe is installed based on the placement points to obtain the probe installation item.

[0025] Based on the probe installation item, a scanning threshold is set, which is a fixed time value. The processing surface is scanned based on the scanning threshold, a 3D point cloud map is reconstructed, and the roughness information is calculated in real time to obtain the real-time accuracy item.

[0026] Furthermore, the method for creating the first association model includes:

[0027] Based on the initial information item, the physical relationship between the initial information item and the real-time accuracy item is extracted to obtain the first extraction item;

[0028] Design a physical information neural network, using initial information terms as the input layer, LSTM to process time-series data as the hidden layer, real-time accuracy terms as the output layer, and the first extracted terms as training data for model training. Output the correlation data between oscillation frequency and feed rate and real-time accuracy terms, and then obtain the first correlation model.

[0029] Furthermore, the method for obtaining the initial adjustment term includes:

[0030] Based on the first association model, with processing requirements as input, and based on the model output, the information combination of the initial information items is obtained to obtain the combined information set;

[0031] Combined damage data, including oscillation frequency damage and feed rate damage, is obtained based on the combined information set. Damage is sorted based on the combined damage data to obtain damage sorting terms. Based on the damage sorting terms, the combination corresponding to the lowest damage is used as the benchmark adjustment combination to obtain the initial adjustment term.

[0032] Furthermore, the adjustment method includes:

[0033] Based on the damage ranking term, a damage range is set, which is a fixed range value. Based on the combination result of the damage range and the lowest damage, the damage fluctuation term is obtained.

[0034] The combination corresponding to the damage fluctuation term is obtained as the secondary adjustment combination. The secondary adjustment combination is used as the adjustment pattern. The update information of vibration information, pressure information and temperature information is obtained respectively. The fluctuation range of the update information is obtained. The combination with the lowest comprehensive fluctuation range is obtained as the adaptation adjustment combination. Then, the initial information term is adjusted in a secondary manner.

[0035] Furthermore, the adjustment method also includes:

[0036] Based on the vibration information item, pressure information item, and temperature information item, respectively set the benchmark weights. Based on the combination result of the fluctuation range of the benchmark weights and the update information, the weight combination item is obtained. The lowest combination among the weight combination items is taken as the update adaptation adjustment combination, and the initial information item is adjusted a second time.

[0037] Compared with the prior art, the beneficial effects of the present invention are:

[0038] This intelligent servo control method for vertical ultra-precision machines based on dynamic parameter identification constructs a four-dimensional sensor network of vibration, pressure, temperature, and morphology by integrating flexible patches of microcavity / thermocouples and a three-band confocal probe. Based on this network, a first correlation model is built. According to the output of the first correlation model, the method prioritizes low-damage combinations of different oscillation frequencies and feed rates for equipment execution by combining damage sorting. Furthermore, it deeply acquires the updated information of vibration, pressure, and temperature information items and performs secondary adjustments on the initial information items, thereby realizing hoist drive control based on dynamic parameter identification of vibration, temperature, and pressure data combinations. Attached Figure Description

[0039] Figure 1 This is a schematic diagram of the overall process of the present invention;

[0040] Figure 2 This is a schematic diagram of the vibration information item acquisition process of the present invention;

[0041] Figure 3 This is a schematic diagram of the process for obtaining processing requirements according to the present invention;

[0042] Figure 4 This is a schematic diagram of the initial adjustment term acquisition process of the present invention;

[0043] Figure 5 This is a schematic diagram of the combined information set of the present invention. Detailed Implementation

[0044] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0045] Long-term operation of vertical ultraprecision machining centers can lead to mechanical wear and vibration during processing, causing precision drift and severely restricting machining quality. Precision drift refers to the deviation between the actual machining position and the theoretical position of the machine during long-term machining due to various factors. This deviation directly leads to a decrease in machining accuracy and may even cause workpiece scrap. The guide rail is a crucial component that supports and guides moving parts. Wear on the guide rail reduces its geometric accuracy, increases the frictional resistance of moving parts, and affects the smoothness and accuracy of their movement. Localized wear on the guide rail can also cause deviations in the movement trajectory of moving parts, thus affecting machining accuracy. In ultra-precision machining, the honing stone comes into direct contact with the workpiece, and its wear directly affects machining accuracy. Wear can alter the honing stone's shape, affecting its contact area and pressure with the workpiece, thus impacting the surface finish and precision. The technical solution provided in this application integrates a flexible patch with a microcavity / thermocouple and a three-band confocal probe to construct a four-dimensional sensing network of vibration, pressure, temperature, and morphology. Based on this network, a first correlation model is built. According to the output of the first correlation model, by combining damage ranking, low-damage combinations of different oscillation frequencies and feed rates are preferentially selected for equipment execution. Furthermore, the updated information of vibration, pressure, and temperature data is acquired at a deeper level, and the initial information is adjusted secondaryly. This achieves honing stone drive control based on dynamic parameter identification of vibration, temperature, and pressure data combinations. Figure 1 As shown, it includes steps S100-S600.

[0046] Step S100: Arrange the first sensing device on the key motion axis of the target device to acquire vibration data of the motion axis and obtain vibration information items.

[0047] It is important to note that, such as Figure 2As shown, the first sensing device is a vibration sensing device. The method for obtaining vibration information items includes: obtaining the motion axis component information of the target vertical ultraprecision machine, including the honing stone supply guide rail, the lead screw, and the C-axis, to obtain installation information items; based on the installation information items, setting an interval value, the interval value being a fixed value, dividing the installation information items into regions based on the interval value, to obtain at least three region division items; arranging vibration sensing devices according to the region division items to obtain vibration data, to obtain a vibration dataset; obtaining the average vibration information of the vibration dataset, and then obtaining the vibration information items.

[0048] Specifically, the set interval value is 200mm. That is, based on the obtained information of the target vertical ultra-precision machine's oilstone supply guide rail, lead screw, and C-axis, the area is divided at 200mm intervals to obtain multiple area items. Vibration sensing devices are arranged in each area item to acquire vibration data, and the average vibration data acquired by each vibration sensing device is calculated as the vibration information item.

[0049] Step S200: Install a second sensing device on the oilstone unit to acquire real-time pressure and temperature data of the oilstone, and obtain pressure information items and temperature information items.

[0050] It should be noted that the second sensing device is an integrated pressure and temperature sensing device, and the second sensing device is a flexible multimodal sensing patch. The method for obtaining the pressure information and temperature information includes: acquiring the working scene information of the oilstone, including the workpiece contact area and the oscillation area; dividing the working scene information of the oilstone into regions based on the workpiece contact area and the oscillation area to obtain a first region and a second region; attaching the flexible sensing patch to the first region and the second region respectively, the flexible sensing patch integrating a microcavity pressure sensor and a thin-film thermocouple sensor to obtain the installation information; and acquiring the pressure data and temperature data based on the installation information to obtain the pressure information and temperature information.

[0051] Specifically, the thickness of the flexible multimodal sensing patch is <0.5mm, which does not interfere with the processing. In the working state of the whetstone, the working scene is divided into regions. The position that contacts the workpiece is determined as the contact area, and the area used to clamp and support the whetstone is determined as the oscillation area. Therefore, the working scene of the whetstone is divided into regions, and flexible sensing patches integrating microcavity pressure sensors and thin-film thermocouple sensors are attached to them to obtain clamping pressure data of the whetstone and temperature data of the oscillation area.

[0052] Step S300: Obtain the processing requirements, which are the precision requirements of the processing surface. Based on the processing requirements, obtain the actual precision data of the target workpiece through the image acquisition device to obtain the real-time precision item.

[0053] It is important to note that, such as Figure 3The method for obtaining processing requirements includes: acquiring processing data of the target workpiece, including processing drawings; performing information scanning based on the processing data to obtain precision annotation data and obtain annotation information items; generating processing parameter instructions based on the annotation information items; using the processing parameter instructions as input to the target vertical ultraprecision machine to obtain output data, thereby obtaining processing requirements.

[0054] Specifically, before starting, the workpiece drawing is scanned, and the precision markings, such as Ra and roundness, are identified by AI. Then, machining parameter instructions are generated, such as conical raceway: Ra≤0.1μm, roundness improvement≥25%, etc. Based on the machining parameter instructions, the target vertical ultraprecision machine is used as input to obtain output data, and thus the machining requirements are obtained.

[0055] It should be noted that the image acquisition device includes a non-contact inspection probe. The method for obtaining the real-time accuracy item includes: the non-contact inspection probe includes a confocal microscopic probe assembly with red-blue-ultraviolet three-band wavelengths, used to penetrate oil mist to obtain nanoscale three-dimensional morphology and acquire processing images; based on the processing images, placement points are set, which are integrated on the side of the oilstone; the non-contact probe is installed based on the placement points, resulting in the probe installation item; based on the probe installation item, a scanning threshold is set, which is a fixed time value; the processing surface is scanned based on the scanning threshold, a 3D point cloud map is reconstructed, and roughness information is calculated in real time, resulting in the real-time accuracy item.

[0056] Specifically, based on the processing image, the contact point between the oilstone and the workpiece in the processing image is used as the judgment point. The placement point is set to the side of the judgment point, and a non-contact probe is installed based on the placement point. Since the non-contact detection probe includes a confocal microscopic probe assembly with red-blue-ultraviolet three-band wavelengths, the spectral range is set to 400-1000nm, the Z-axis resolution to 0.05μm, and the set scanning threshold to 5s. Based on the scanning threshold, the non-contact detection probe continuously scans the processing surface, with the scanning path synchronized. The probe is linked to the oilstone via a ball screw to ensure that the scanning trajectory covers the processing area. The imaging process is as follows: In the ultra-precision cycle step, after each section of raceway processing is completed, the probe triggers a high-speed scan. The scanning time is less than 0.5 seconds per frame, meeting the cycle time requirement. Data is acquired through multispectral channels. The topography layer involves scanning the surface with 6 wavelength beams to generate 3D point clouds with a resolution of 0.1μm. The material layer analyzes the differences in reflectance spectra to identify residual cutting fluid and oilstone material adhesion. Then, an AI requirement analysis engine is built. The data processing flow includes defect classification. Image segmentation involves AI identifying raceways and edge areas to eliminate interference from non-machined areas. Defect classification involves training a CNN model to identify 7 typical defects, such as scratches, dents, and adhesive wear, and associating them with oilstone parameters. The data is then output through AI and fused with a physical model to calculate real-time accuracy, thus obtaining the real-time accuracy item.

[0057] Step S400: Obtain initial data of the oilstone, including oscillation frequency and feed rate, to obtain initial information items. Based on the difference between the real-time accuracy item and the processing requirements, create a first correlation model based on the initial information.

[0058] It should be noted that the first correlation model outputs correlation data between different initial data and machining surface accuracy data. The creation method of the first correlation model includes: based on the initial information item, extracting the physical relationship between the initial information item and the real-time accuracy item to obtain the first extraction item; designing a physical information neural network, using the initial information item as the input layer, LSTM processing time series data as the hidden layer, the real-time accuracy item as the output layer, and the first extraction item as the training data for model training, outputting the correlation data between the oscillation frequency and the feed rate and the real-time accuracy item, thereby obtaining the first correlation model.

[0059] Specifically, the physical relationship between the initial information item and the real-time accuracy item is extracted, including the physical relationship between roughness Ra and oscillation frequency f, to obtain the first extracted item. A physical information neural network is designed, whose model architecture includes sensor data, physical constraint layer, neural network backbone, optimized control instructions, and equipment execution. The fusion core includes embedding physical relationship equations and outputs that satisfy physical laws. The physical constraint equations include the guide rail wear model and the oilstone fatigue accumulation model, where the guide rail wear model is:

[0060] ;

[0061] in This is due to the guide rail positioning error. This is the wear rate coefficient. For sliding friction, For feed rate, For material activation energy, Let be the ideal gas constant. The temperature of the workbench;

[0062] The oilstone fatigue accumulation model is as follows:

[0063] ;

[0064] in This represents the cumulative fatigue value of the oilstone. This represents the amplitude of the contact stress of the asphalt stone. This represents the fatigue limit stress of the oilstone. The fatigue index of the material. The oscillation frequency is then used to construct the neural network backbone:

[0065] import torch

[0066] import torch.nn as nn

[0067] class PINN_Model(nn.Module):

[0068] def __init__(self):

[0069] super().__init__()

[0070] # Sensor data input layer (4-dimensional: vibration, temperature, pressure, frequency)

[0071] self.input_layer = nn.Linear(4, 64)

[0072] # Physical Information Embedding Layer

[0073] self.physics_constraint = nn.Sequential(

[0074] nn.Linear(64, 32),

[0075] PhysicsActivation() # Custom activation function to force the execution of the equation in step 1. )

[0077] # Dual-task output

[0078] self.task_branch = nn.ModuleDict({

[0079] "wear_comp": nn.Linear(32, 1), # Output wear compensation amount Δx

[0080] "life_pred": nn.Linear(32, 1) # Outputs the remaining lifespan of the oilstone.

[0081] })

[0082] def forward(self, x):

[0083] x = torch.relu(self.input_layer(x))

[0084] x = self.physics_constraint(x) # Physics constraints take effect

[0085] return {

[0086] "compensation": self.task_branch["wear_comp"](x),

[0087] "life_left": torch.sigmoid(self.task_branch["life_pred"](x))

[0088] }

[0089] # Physical activation function (forces satisfaction of differential equations)

[0090] class PhysicsActivation(nn.Module):

[0091] def forward(self, x):

[0092] # The wear model differential equation from step 1 is embedded here.

[0093] dxdt = k1×F_f× v× torch.exp(-Q / (R×T))

[0094] return x× dxdt # Physical laws as constraints

[0095] Simultaneously, loss functions are designed, including data fitting loss and physical law loss, which include constraint one and constraint two. Constraint one: the rate of change of positioning error must satisfy the wear model. Constraint two: the probability of fracture when the oilstone life is exhausted is >99%. Then, real-time control closed loop and model output are performed to obtain the first correlation model.

[0096] Step S500: Based on the processing requirements as input, the initial information items are initially adjusted through the first association model to obtain the initial adjustment items.

[0097] It is important to note that, such as Figure 4 As shown, the method for obtaining the initial adjustment term includes: based on the first correlation model, taking the processing requirements as input, and based on the model output results, obtaining the information combination of the initial information terms to obtain the combined information set; based on the combined information set, obtaining the combined damage data, including oscillation frequency damage and feed speed damage; based on the combined damage data, sorting the damage to obtain the damage sorting term; based on the damage sorting term, taking the combination corresponding to the lowest damage as the benchmark adjustment combination, and thus obtaining the initial adjustment term.

[0098] Specifically, such as Figure 5As shown, under the premise of fixed processing requirements, multiple combinations of initial information items can be obtained based on the model's output. For example, when the surface roughness Ra of the arc does not exceed 0.15μm, it can be achieved through multiple combinations of different oscillation frequencies and feed rates, including combination 1: oscillation frequency A, feed rate B; combination 2: oscillation frequency C, feed rate D; combination 3: oscillation frequency E, feed rate F, etc. Different oscillation frequencies and feed rates will cause different damage to the machine. At this time, the combined damage data is obtained. For example, the damage caused by oscillation frequencies A, C, and E is 2%, 3%, and 3%, respectively, and the damage caused by feed rates B, D, and F is 4%, 2%, and 1%, respectively. At this time, the combined damage data of combination 1, combination 2, and combination 3 are 6%, 5%, and 4%, respectively. Then, the damage is sorted based on the combined damage data to obtain the damage sorting item, that is, combination 3 < combination 2 < combination 1. Based on the damage sorting item, the combination corresponding to the lowest damage is used as the benchmark adjustment combination, and then the initial adjustment item is obtained, that is, the initial information item is adjusted with oscillation frequency E and feed rate F.

[0099] Step S600: Based on the initial adjustment terms, obtain the updated information of vibration information terms, pressure information terms, and temperature information terms, and perform secondary adjustment on the initial information terms through the adjustment method.

[0100] It is important to note that secondary adjustment of the initial information items enables oilstone drive control based on dynamic parameter identification of combinations of vibration, temperature, and pressure data. The adjustment method includes: setting a damage range based on the damage ranking item, with the damage range being a fixed range value; obtaining the damage fluctuation item based on the combination result of the damage range and the lowest damage; obtaining the combination corresponding to the damage fluctuation item as the secondary adjustment combination; using the secondary adjustment combination as the adjustment pattern; obtaining the update information of the vibration, pressure, and temperature information items respectively; obtaining the fluctuation range of the update information; obtaining the combination with the lowest comprehensive fluctuation range as the adaptive adjustment combination; and then performing secondary adjustment of the initial information items.

[0101] Specifically, the damage range is set to 50%. Based on the combination result of the damage range and the minimum damage, the damage fluctuation term is obtained. When the combination information is: Combination 1: oscillation frequency A, feed speed B; Combination 2: oscillation frequency C, feed speed D; Combination 3: oscillation frequency E, feed speed F, etc., different oscillation frequencies and feed speeds will cause different damage to the machine. At this time, the combined damage data is obtained. For example, the damage caused by oscillation frequencies A, C, and E is 2%, 3%, and 3% respectively, and the damage caused by feed speeds B, D, and F is 4%, 2%, and 1% respectively. At this time, the combined damage data of Combination 1, Combination 2, and Combination 3 are 6%, 5%, and 4% respectively. When the combined result of the minimum damage is 4%, the damage fluctuation term is 6%. The combination corresponding to the damage fluctuation term is obtained as the secondary adjustment combination, namely Combination 2 and Combination 3. Using the secondary adjustment combination as the adjustment mode, the update information of vibration information, pressure information, and temperature information is obtained respectively, and the fluctuation range of the update information is obtained. The obtained data is shown in Table 1.

[0102] Table 1

[0103]

[0104] The combination with the lowest comprehensive fluctuation range is obtained as the adaptation adjustment combination, namely the oscillation frequency C and feed speed D corresponding to combination 2, and then the initial information items are adjusted a second time.

[0105] It should be noted that the adjustment method also includes: setting benchmark weights based on vibration information items, pressure information items, and temperature information items respectively; obtaining weight combination items based on the combination results of the fluctuation range of the benchmark weights and the updated information; taking the lowest combination among the weight combination items as the update adaptation adjustment combination; and performing secondary adjustment on the initial information items.

[0106] Specifically, the impact of vibration, pressure, and temperature information varies across different devices. For example, due to equipment aging, the impact of vibration information may be far greater than that of pressure and temperature information. In this case, the weights are adjusted, such as setting the weight of vibration information to 60%, and pressure and temperature information to 20% each. Then, the adjusted weights are combined with the updated information to obtain the lowest possible combination as the update adaptation adjustment combination, and the initial information items are adjusted a second time.

[0107] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended embodiments and their equivalents.

Claims

1. A method for intelligent servo control of a vertical ultraprecision machine based on dynamic parameter identification, comprising: A first sensing device is arranged on the key motion axis of the target device to acquire vibration data of the motion axis and obtain vibration information items. Its characteristic is that it further includes: A second sensing device is installed on the oilstone unit. The second sensing device is an integrated pressure and temperature sensing device, which is used to acquire real-time pressure data and temperature data of the oilstone to obtain pressure information items and temperature information items. The processing requirements are obtained, which are the accuracy requirements of the processing surface. Based on the processing requirements, the actual accuracy data of the target workpiece is obtained through image acquisition equipment to obtain the real-time accuracy item. Acquire initial data of the oilstone, including oscillation frequency and feed rate, to obtain initial information items. Based on the difference between the real-time accuracy item and the processing requirements, create a first correlation model based on the initial information. The first correlation model outputs correlation data between different initial data and processing surface accuracy data. Based on the processing requirements as input, perform preliminary adjustment on the initial information items through the first correlation model to obtain initial adjustment items. Based on the initial adjustment term, the updated information of vibration information, pressure information and temperature information is obtained. The initial information term is then adjusted a second time through the adjustment method, thereby realizing the oilstone drive control based on the dynamic parameter identification of the combination of vibration data, temperature data and pressure data. The method for obtaining the initial adjustment term includes: Based on the first association model, with processing requirements as input, and based on the model output, the information combination of the initial information items is obtained to obtain the combined information set; Combined damage data, including oscillation frequency damage and feed rate damage, is obtained based on the combined information set. Damage is sorted based on the combined damage data to obtain damage sorting terms. Based on the damage sorting terms, the combination corresponding to the lowest damage is used as the benchmark adjustment combination to obtain the initial adjustment term.

2. The intelligent servo control method for a vertical ultraprecision machine based on dynamic parameter identification according to claim 1, characterized in that: The first sensing device is a vibration sensing device, and the method for acquiring vibration information items includes: Obtain information on the motion axis components of the target vertical ultraprecision machine, including the oilstone supply guide rail, lead screw, and C-axis, and obtain the installation information items; Based on the installation information items, an interval value is set. The interval value is a fixed value. The installation information items are divided into regions based on the interval value to obtain at least three region division items. Vibration sensing devices are arranged according to the region division items to obtain vibration data and obtain a vibration dataset. The average vibration information of the vibration dataset is obtained, and then the vibration information item is obtained.

3. The intelligent servo control method for a vertical ultraprecision machine based on dynamic parameter identification according to claim 1, characterized in that: The second sensing device is a flexible multimodal sensing patch, and the methods for acquiring pressure and temperature information items include: Acquire oilstone working scene information, including workpiece contact area and oscillation area, and divide the oilstone working scene information into regions based on workpiece contact area and oscillation area to obtain first region and second region; Based on the first region and the second region, flexible sensing patches are attached respectively. The flexible sensing patches integrate a microcavity pressure sensor and a thin-film thermocouple sensor to obtain installation information items. Based on the installation information items, pressure data and temperature data are obtained, resulting in pressure information items and temperature information items.

4. The intelligent servo control method for a vertical ultraprecision machine based on dynamic parameter identification according to claim 1, characterized in that: The methods for obtaining the processing requirements include: Acquire the target workpiece machining data, including machining drawings; perform information scanning based on the machining data to obtain precision annotation data and obtain annotation information items; Based on the labeled information items, processing parameter instructions are generated. These processing parameter instructions are then used as input to the target vertical ultraprecision machine to obtain output data, thereby obtaining the processing requirements.

5. The intelligent servo control method for a vertical ultraprecision machine based on dynamic parameter identification according to claim 1, characterized in that: The image acquisition device includes a non-contact detection probe, and the method for obtaining the real-time accuracy item includes: The non-contact detection probe includes a confocal microscopic probe assembly with red-blue-ultraviolet three-bands, which is used to penetrate oil mist to obtain nanoscale three-dimensional morphology and acquire processing images. Based on the processing images, placement points are set, which are integrated on the side of the oilstone. The non-contact probe is installed based on the placement points to obtain the probe installation item. Based on the probe installation item, a scanning threshold is set, which is a fixed time value. The processing surface is scanned based on the scanning threshold, a 3D point cloud map is reconstructed, and the roughness information is calculated in real time to obtain the real-time accuracy item.

6. The intelligent servo control method for a vertical ultraprecision machine based on dynamic parameter identification according to claim 1, characterized in that: The method for creating the first association model includes: Based on the initial information item, the physical relationship between the initial information item and the real-time accuracy item is extracted to obtain the first extraction item; Design a physical information neural network, using initial information terms as the input layer, LSTM to process time-series data as the hidden layer, real-time accuracy terms as the output layer, and the first extracted terms as training data for model training. Output the correlation data between oscillation frequency and feed rate and real-time accuracy terms, and then obtain the first correlation model.

7. The intelligent servo control method for a vertical ultraprecision machine based on dynamic parameter identification according to claim 1, characterized in that: The adjustment method includes: Based on the damage ranking term, a damage range is set, which is a fixed range value. Based on the combination result of the damage range and the lowest damage, the damage fluctuation term is obtained. The combination corresponding to the damage fluctuation term is obtained as the secondary adjustment combination. The secondary adjustment combination is used as the adjustment pattern. The update information of vibration information, pressure information and temperature information is obtained respectively. The fluctuation range of the update information is obtained. The combination with the lowest comprehensive fluctuation range is obtained as the adaptation adjustment combination. Then, the initial information term is adjusted in a secondary manner.

8. The intelligent servo control method for a vertical ultraprecision machine based on dynamic parameter identification according to claim 7, characterized in that: The adjustment method further includes: Based on the vibration information item, pressure information item, and temperature information item, respectively set the benchmark weights. Based on the combination result of the fluctuation range of the benchmark weights and the update information, the weight combination item is obtained. The lowest combination among the weight combination items is taken as the update adaptation adjustment combination, and the initial information item is adjusted a second time.

Images