Intelligent regulation and control method and system for clamping force of skylight guide rail clamp

Abstract

The invention relates to the technical field of intelligent manufacturing and precision machining, and discloses a skylight guide rail clamp clamping force intelligent regulation and control method and system.The skylight guide rail clamp clamping force intelligent regulation and control method comprises the steps that multi-dimensional sensing data of a skylight guide rail clamp is obtained and preprocessed; acquiring a force displacement data pair and identifying a flexibility coefficient of the guide rail; the elasticity modulus is inversely calculated through the flexibility coefficient, and temperature correction is carried out; calculating theoretical stress distribution and a displacement field, and fusing and correcting the theoretical stress distribution and the displacement field with measured data of the sensor; evaluating stress safety, deformation stability and vibration stability; the current clamping force distribution is adjusted; the clamping force adjustment amount is calculated and converted into a control instruction for adjustment, the guide rail is machined, and quality detection is conducted; updating the quality prediction model, predicting tool wear and calculating compensation; according to the invention, through an online learning and adaptive optimization mechanism, the system can continuously learn and evolve and automatically adapt to changes of production conditions.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent manufacturing and precision machining, more particularly, it relates to a sunroof guide rail clamp clamping force intelligent regulation method and system. BACKGROUND

[0002] In the field of automobile sunroof manufacturing, the sunroof guide rail, as the core sliding component of the sunroof system, its machining precision and surface quality directly affect the opening and closing smoothness, sealing performance and overall service life of the sunroof. The sunroof guide rail is usually made of aluminum alloy extruded profile and needs to go through multiple machining processes such as precision milling, drilling and tapping. In these machining processes, the clamping force regulation of the guide rail clamp is crucial. Too small clamping force will cause vibration and displacement of the workpiece during high-speed cutting, seriously affecting the machining precision and surface quality, and too large clamping force will cause elastic-plastic deformation of the thin-walled aluminum alloy guide rail, and even springback deformation after the clamping is removed, which cannot meet the requirements of high-precision machining.

[0003] Currently, the traditional clamp clamping force regulation mainly relies on manual adjustment by the experience of the operator or adjustment by the fixed pressure pneumatic or hydraulic clamping system. This method has many limitations: it cannot be dynamically adjusted in real time according to the changes of the material properties of different batches of guide rails, machining process parameters and tool wear state; it is difficult to monitor the actual stress state and deformation of the workpiece during machining, and problems such as insufficient or excessive clamping force are likely to occur. The existing technology mainly has the following deficiencies: the closed-loop control based on a single pressure sensor cannot comprehensively reflect the stress distribution state of the workpiece under multi-point clamping conditions, and local stress concentration is likely to occur; the threshold-based alarm mechanism is a passive control method and cannot predict and actively regulate before deformation occurs; the offline optimization method based on finite element simulation is difficult to adapt to the influence of various uncertain factors such as material property fluctuations, tool wear and environmental temperature changes in the actual machining process.

[0004] Therefore, it is urgent to provide a sunroof guide rail clamp clamping force intelligent regulation method and system, which can realize dynamic perception, intelligent decision-making and precise regulation of the clamping force during machining, solve the clamping force regulation problem in the machining process of thin-walled guide rails, change from passive control to active prediction and from experience adjustment to intelligent decision-making, improve machining precision, surface quality and production efficiency, reduce workpiece rejection rate and energy consumption, and provide an innovative solution for the intelligent development of the precision manufacturing field. SUMMARY

[0005] The present application provides a sunroof guide rail clamp clamping force intelligent regulation method and system, which solves the technical problems in the related art that the clamping force of the thin-walled guide rail clamp cannot be intelligently regulated, it is difficult to adapt to material property fluctuations and changes in machining conditions, and it leads to low machining precision and high workpiece rejection rate.

[0006] This invention provides a method for intelligent control of the clamping force of a sunroof guide rail clamp, comprising: The multi-dimensional sensing data of the sunroof guide rail fixture is acquired and preprocessed to obtain a multi-source synchronous sensing dataset, including displacement sensing data, temperature sensing data, current clamping force distribution and vibration sensing data. The force-displacement data pairs were obtained by using a micro-loading test method on the displacement sensing data and the compliance coefficient of the guide rail was identified. The elastic modulus was calculated by back-calculating the compliance coefficient and temperature correction was performed by combining it with temperature sensing data to obtain the actual material property parameters. Based on actual material properties and current clamping force distribution, theoretical stress distribution and displacement field are calculated and fused with sensor measured data for correction to obtain the stress and deformation state analysis results of the guide rail. Based on the analysis results of guide rail stress and deformation state and vibration sensing data, stress safety, deformation stability and vibration stability are evaluated, and a safety level is generated. Adjust the current clamping force distribution according to the safety level to obtain the clamping force configuration result; The clamping force adjustment amount is calculated based on the clamping force configuration result and converted into control commands for adjustment, thus obtaining the clamping force distribution result; the guide rail is machined based on the clamping force distribution result and quality inspection is performed to obtain the machining quality inspection result; The quality prediction model is updated based on the clamping force distribution results and machining quality inspection results. Tool wear is predicted and compensation is calculated to obtain a clamping force control scheme.

[0007] In a preferred embodiment, the acquisition of the multi-dimensional sensing data includes: Obtain the physical structure information of the sunroof guide rail clamping system, and install piezoresistive pressure sensors at each clamping point based on the structural characteristics of the six-point clamping fixture to measure the normal force applied at the clamping point in real time. Laser displacement sensors are deployed at key locations on the guide rail to detect minute deformations of the guide rail; A triaxial accelerometer is installed at the center of the fixture body to monitor the vibration characteristics during the machining process; Infrared temperature sensors are placed above the working area to monitor the temperature distribution on the guide rail surface and obtain multi-dimensional physical quantity data.

[0008] In a preferred embodiment, obtaining the actual material property parameters includes: The preprocessed multi-source synchronous sensing dataset is acquired, and the clamping system is controlled to apply a preset small-amplitude clamping force sequence to the preprocessed multi-source synchronous sensing dataset. The deformation response of the displacement measuring point is collected synchronously to obtain force-displacement data pairs. Based on force and displacement data, a linear relationship model is established using the least squares method. The flexibility coefficient of the linear relationship model is then identified. The identified flexibility coefficient is then back-calculated using the guide rail geometric parameters to obtain the actual elastic modulus. The actual elastic modulus and yield strength are corrected based on the average temperature of the guide rail measured by the temperature sensor.

[0009] In a preferred embodiment, the acquisition of the guide rail stress and deformation state analysis results includes: The actual material properties and current clamping force distribution are obtained, and the guide rail is discretized using the Timoshenko beam model, which is then discretized into beam elements. Elastic constraint boundary conditions are applied to the beam elements at the six clamping points, and cutting loads are applied to the elements corresponding to the current tool position. The modal superposition method is used to solve the beam element after applying load with reduced order, and the theoretical stress distribution and displacement field are calculated. The theoretical calculation results are fused and corrected with the sensor's measured data.

[0010] In a preferred embodiment, the generation of the security level includes: Obtain the stress and deformation state analysis results of the guide rail and the measured vibration data. Calculate the stress ratio based on the maximum Von Mises stress and the actual yield strength of the material in the analysis results, and assess the stress safety. The deformation stability is evaluated by comparing the maximum displacement in the analysis results with the preset deformation threshold. Based on the analysis of measured vibration data, vibration amplitude and spectral characteristics are analyzed to assess vibration stability. The safety level is generated by combining the assessment results of stress safety, deformation stability, and vibration stability.

[0011] In a preferred embodiment, the solution for the clamping force configuration result includes: Obtain safety assessment results, and when the safety level is warning or danger, establish a multi-objective optimization model based on the assessment results with the objectives of minimizing stress, minimizing deformation, and suppressing vibration. The particle swarm optimization algorithm is used to solve the optimization model, and the optimization efficiency is improved by using a hot start strategy and adaptive inertia weight. The optimal solutions are sorted by multi-objectives based on the Pareto dominance relationship, and the optimal solution is selected as the clamping force configuration scheme.

[0012] In a preferred embodiment, obtaining the clamping force distribution result includes: Obtain the optimal clamping force configuration result, calculate the clamping force adjustment amount at each clamping point based on the configuration result, and convert the adjustment amount into servo motor control commands; An S-shaped speed curve is used to smoothly adjust and control the servo motor, and a force closed-loop PID control is used to precisely adjust the clamping force. The effect of the adjusted clamping force was verified, and the data of the adjustment process and the final clamping force distribution were recorded.

[0013] In a preferred embodiment, obtaining the clamping force control scheme includes: Obtain historical adjustment data and processing quality results, and update the parameters of the quality prediction model based on the historical data using an incremental learning algorithm; The quality prediction model was rebuilt and trained using support vector regression. The updated model is evaluated for performance, and the algorithm parameters of the control strategy are optimized based on the evaluation results. Retrain the model periodically to maintain prediction accuracy.

[0014] In a preferred embodiment, the prediction of tool wear and calculation of compensation includes: Acquire cutting force variation data during the cutting process, and monitor the trend of the variation data to identify tool wear status; An empirical model of tool wear and cutting force increment is established based on tool wear status; Predict the incremental cutting force based on the empirical model according to the cumulative tool usage time; The clamping force compensation required to cope with wear is calculated based on the prediction results, and a tool replacement prompt is provided when the wear reaches a preset threshold.

[0015] In a preferred embodiment, a sunroof guide rail clamping force intelligent control system is used to execute the above-described sunroof guide rail clamping force intelligent control method, including: The sensor data acquisition module is used to acquire multi-dimensional sensor data of the sunroof guide rail fixture and preprocess it to obtain a multi-source synchronous sensor dataset, including displacement sensor data, temperature sensor data, current clamping force distribution and vibration sensor data. The material property acquisition module is used to obtain force-displacement data pairs from displacement sensing data using a micro-loading test method and identify the compliance coefficient of the guide rail; the elastic modulus is calculated from the compliance coefficient and temperature correction is performed by combining it with temperature sensing data to obtain the actual material property parameters. The state analysis and calculation module is used to calculate the theoretical stress distribution and displacement field based on the actual material property parameters and the current clamping force distribution, and to fuse and correct them with the sensor measured data to obtain the stress and deformation state analysis results of the guide rail. The safety assessment module is used to assess stress safety, deformation stability, and vibration stability based on the analysis results of guide rail stress and deformation state and vibration sensing data, and to generate a safety level. The clamping force configuration module is used to adjust the current clamping force distribution according to the safety level and obtain the clamping force configuration result. The execution adjustment control module is used to calculate the clamping force adjustment amount based on the clamping force configuration result and convert it into control commands for adjustment, thereby obtaining the clamping force distribution result; based on the clamping force distribution result, the guide rail is machined and quality inspection is performed to obtain the machining quality inspection result; The adaptive learning module is used to update the quality prediction model based on the clamping force distribution results and machining quality inspection results, predict tool wear and calculate compensation, and obtain a clamping force control scheme.

[0016] The beneficial effects of this invention are as follows: By integrating multi-dimensional sensor networks, digital twin modeling, multi-objective optimization decision-making, and precise execution control, a complete intelligent control closed loop is constructed, achieving a technological leap from passive response to active prediction; compared with the traditional fixed pressure clamping method, this invention can comprehensively perceive the complex mechanical state of the guide rail under multi-point clamping and dynamic cutting loads, and predict stress distribution and deformation trends in real time through the digital twin model, making preventive adjustments before problems occur, thus avoiding the lag problem caused by passive control; at the same time, the multi-objective optimization algorithm can achieve intelligent balance among multiple performance indicators such as accuracy, stability, and energy consumption, and automatically select the optimal control strategy according to the priority of specific processing tasks; Through online learning and adaptive optimization mechanisms, the system can continuously learn and evolve, automatically adapting to changes in production conditions. Incremental learning algorithms continuously update the quality prediction model based on actual machining quality feedback, and reverse-optimize and adjust algorithm parameters, continuously improving the system's control strategies. Tool wear tracking and predictive compensation mechanisms can identify the impact of tool wear on cutting forces in advance, adjusting clamping forces to cope with changes in working conditions and avoiding machining quality degradation caused by tool wear. Working condition space coverage assessment and proactive exploration strategies ensure the system accumulates comprehensive operational experience, enhancing its generalization ability. The system maintained stable control performance during three consecutive months of production, successfully adapting to the differences in characteristics of five batches of materials from different suppliers and seasonal environmental temperature fluctuations, demonstrating good robustness and adaptability. Compared to traditional methods that require frequent manual parameter adjustments, this invention achieves truly intelligent and unmanned production, reducing reliance on operator experience, improving production efficiency and quality consistency, and aligning with the development direction of modern intelligent manufacturing. Attached Figure Description

[0017] Figure 1 This is a flowchart of a method for intelligent control of the clamping force of a sunroof guide rail clamp in this invention; Figure 2 This is a module diagram of an intelligent control system for the clamping force of a sunroof guide rail clamp in this invention. Detailed Implementation

[0018] The subject matter described herein will now be discussed with reference to exemplary embodiments. It should be understood that these embodiments are discussed only to enable those skilled in the art to better understand and implement the subject matter described herein, and changes may be made to the function and arrangement of the elements discussed without departing from the scope of this specification. Various processes or components may be omitted, substituted, or added as needed in the examples. Furthermore, some features described in the examples may be combined in other examples.

[0019] At least one embodiment of the present invention discloses an intelligent control method for the clamping force of a sunroof guide rail clamp, such as... Figure 1 As shown, it includes: Step 1: Acquire multi-dimensional sensing data of the sunroof guide rail fixture and preprocess it to obtain a multi-source synchronous sensing dataset, including displacement sensing data, temperature sensing data, current clamping force distribution and vibration sensing data. This step is based on the physical structure of the sunroof guide rail clamping system. It uses a multi-type sensor network to collect multi-dimensional physical quantities in real time, and achieves high-precision time alignment through hardware clock synchronization and software timestamp correction methods to obtain a reliable multi-source synchronous sensor dataset.

[0020] The sensor network deployment is specifically as follows: Based on the structural characteristics of the six-point clamping fixture, a piezoresistive pressure sensor is installed at each clamping point, with a measurement range of 0 to 3000 N and a measurement accuracy of 0.5% of full scale. The pressure sensor is embedded in the clamping contact surface to directly measure the normal force applied to the workpiece by the clamping point. Laser displacement sensors are deployed at key locations on the guide rail, totaling eight measuring points located at both ends, the midpoint, and between each clamping point, with a measurement resolution of 1 μm, to detect minute deformations of the guide rail. A triaxial accelerometer is installed at the center of the fixture body, with a measurement range of ±50g, to monitor vibration characteristics during machining. A non-contact infrared temperature sensor is arranged above the working area, with a measurement range of 0 to 150 degrees Celsius and an accuracy of ±0.5 degrees Celsius, to monitor the temperature distribution on the guide rail surface. All sensors are connected to the data acquisition system via shielded cables to reduce electromagnetic interference.

[0021] Hardware clock synchronization sampling is implemented by using a multi-channel data acquisition card with a unified master clock. The master clock uses a temperature-compensated crystal oscillator with a frequency stability better than ±10ppm. Signals from pressure, displacement, and acceleration sensors are input to different channels of the data acquisition card, with a sampling frequency set to 1000Hz. The data acquisition card internally generates a unified sampling trigger signal to synchronously trigger analog-to-digital conversion across all channels, ensuring strict consistency in the sampling timing of data from each channel, with a hardware-level time deviation of less than 1 microsecond. For the infrared temperature sensor, which uses a serial digital interface, hardware synchronization is not possible; therefore, asynchronous sampling is performed separately at a frequency of 100Hz.

[0022] Software timestamp correction and data alignment are implemented, specifically as follows: For asynchronously sampled temperature sensor data, a software timestamp correction method is used to achieve time alignment with other sensors. In the data acquisition program, the acquisition time and arrival time of each temperature data packet are recorded, and the average time delay of the temperature sensor is obtained through multiple measurements and statistics. Delay compensation correction is applied to the timestamps of historical temperature data. A linear interpolation method is used to interpolate the corrected temperature data to the same time base as other sensors, achieving time alignment of data from various sensors. A time window width of 10 milliseconds is set, and the sensor data arriving within the time window are combined into a synchronized data frame for subsequent processing.

[0023] Sensor signal preprocessing is performed, specifically: The raw sensor signals acquired synchronously are preprocessed to eliminate noise and outliers. For pressure and displacement sensor signals, a sliding window averaging filter is used, with the window length adaptively determined based on the signal's spectral characteristics. The power spectral density of the signal is calculated, identifying the main frequency components and noise frequency components. The window length is selected so that its corresponding cutoff frequency lies between the main signal frequency and the noise frequency, preserving useful signals while suppressing high-frequency noise. For accelerometer signals, a high-pass filter is used to eliminate low-frequency drift, with the cutoff frequency set to 10Hz. Data cross-validation is performed using the spatial correlation between sensors: when the reading of a pressure sensor deviates from that of adjacent sensors and does not conform to physical laws, it is marked as suspicious data and corrected using Kalman filtering based on the statistical characteristics of historical data. Unit normalization is performed on all types of sensor data, mapping different physical quantities to a dimensionless space to facilitate subsequent fusion analysis.

[0024] After the above processing, a preprocessed multi-source synchronous sensing dataset is obtained, including real-time pressure values ​​at six clamping points, displacement values ​​at eight measuring points, triaxial vibration acceleration, and temperature distribution on the guide rail surface. The data sampling frequency is 1000Hz, the time synchronization accuracy is better than 1 microsecond, and the signal-to-noise ratio is improved.

[0025] Step 2: Use a micro-loading test method to obtain force-displacement data pairs from the displacement sensing data and identify the compliance coefficient of the guide rail; calculate the elastic modulus from the compliance coefficient and perform temperature correction by combining it with temperature sensing data to obtain the actual material property parameters. This step, based on the preprocessed multi-source synchronous sensor dataset, uses micro-loading experiments and parameter identification methods to obtain the actual material property parameters of the current batch of guide rails, including the elastic modulus and the temperature-corrected yield strength.

[0026] The design of the micro-loading test involves performing a material property identification test before formal machining begins. A pre-designed sequence of small clamping forces is applied to the control clamping system. This sequence comprises five different force levels, increasing progressively from 300 N to 700 N, with each level held for 2 seconds. These force levels are significantly lower than the material's yield load, ensuring that the guide rail only undergoes elastic deformation. Simultaneously, the deformation response at eight displacement measurement points is acquired while the clamping force is applied. Since the guide rail is a beam structure, the deformation under multi-point clamping force can be calculated using the superposition principle. The steady-state displacement value corresponding to each force level is recorded, obtaining force-displacement data pairs.

[0027] The compliance coefficient is identified, specifically by using the least squares method to identify the compliance coefficient of the guide rail based on the acquired force-displacement data pairs. A linear relationship model is established, expressing the displacement as a linear function of the clamping force, with the compliance coefficient representing the slope. For beam structures, the relationship between the midpoint displacement and the concentrated load is determined through the following calculation steps: Obtain the applied concentrated load value, obtain the span length of the beam, then obtain the elastic modulus value of the material, and then obtain the moment of inertia value of the section. Multiply the load value by the cube of the span length, divide the product by 48, divide the result by the elastic modulus, and finally divide the result by the moment of inertia of the section to obtain the displacement of the beam midpoint.

[0028] The units for the displacement at the midpoint of the beam are millimeters, the units for the applied concentrated load are Newtons, the units for the span of the beam are millimeters, the units for the elastic modulus of the material are megapascals, and the units for the moment of inertia of the cross section are millimeters to the fourth power.

[0029] For complex boundary conditions involving multi-point clamping, the influence coefficient matrix method is used to establish the relationship between displacement and force. A compliance influence coefficient matrix is ​​defined, where the element in the i-th row and j-th column represents the displacement generated at the i-th measuring point when a unit force is applied at the j-th clamping point.

[0030] The compliance matrix is ​​inversely proportional to the elastic modulus. The elements of the compliance matrix are fitted using the least squares method based on measured force-displacement data.

[0031] The calculation of the actual elastic modulus involves: based on the identified compliance matrix and combined with the geometric parameters of the guide rail, the actual elastic modulus is calculated in reverse. The guide rail cross-section is a complex profile, and the moment of inertia of the cross-section is pre-calculated using finite element software. The identified compliance coefficients are compared with the theoretical compliance formula to obtain the actual elastic modulus.

[0032] Considering the impact of measurement noise, the identification results of multiple measuring points and multiple loading conditions are averaged to obtain a robust estimate of the elastic modulus.

[0033] Temperature correction and yield strength adjustment are performed, specifically by adjusting the material properties based on the average temperature of the guide rail measured by temperature sensors. The elastic modulus and yield strength of aluminum alloys decrease with increasing temperature. The average temperature of the guide rail is calculated using the arithmetic mean of temperature values ​​from multiple measuring points; that is, the values ​​from all temperature sensors are added together and then divided by the number of sensors to obtain the average temperature representing the overall temperature state of the guide rail.

[0034] The elastic modulus is corrected using an empirical temperature correction formula. The specific calculation steps are as follows: To obtain the elastic modulus at the reference temperature, calculate its temperature coefficient. Then, obtain the average temperature of the guide rail and the reference temperature, calculate the temperature difference between them, multiply the temperature difference by the temperature coefficient, subtract the product from 1, and finally multiply the reference elastic modulus by the result to obtain the temperature-corrected actual elastic modulus. The typical temperature coefficient of the elastic modulus for 6063 aluminum alloy is 0.0003 degrees Celsius to the power of negative 1. The average temperature of the guide rail is in degrees Celsius, and the reference temperature is typically taken as 20 degrees Celsius.

[0035] Similarly, the temperature correction for yield strength is calculated using the following steps: Obtain the nominal yield strength value at the reference temperature, obtain the temperature coefficient of the yield strength, then obtain the average temperature of the guide rail and the reference temperature, calculate the temperature difference between the two, multiply the temperature difference by the temperature coefficient, subtract the product from 1, and finally multiply the nominal yield strength by the result to obtain the actual yield strength after temperature correction.

[0036] After the above processing, the actual material property parameters of the current batch of guide rails are obtained, including the actual elastic modulus and the temperature-corrected yield strength. These parameters truly reflect the mechanical properties of the current guide rails, providing accurate material input for subsequent digital twin modeling.

[0037] Step 3: Based on the actual material properties and the current clamping force distribution, calculate the theoretical stress distribution and displacement field, and fuse and correct them with the sensor measured data to obtain the analysis results of the guide rail stress and deformation state. Based on the identified actual material parameters, the current clamping force distribution, and the estimated cutting force, this step uses the reduced-order finite element method to construct a digital twin model of the guide rail clamp coupling system, thereby obtaining real-time stress and deformation state predictions for the guide rail.

[0038] The guide rail beam model is discretized, specifically by simplifying it into a Timoshenko beam model. This model considers both bending and shear deformation and is suitable for short, thick beams and thin-walled structures. It is discretized into 50 beam elements along the length of the guide rail, each element being 24 mm long. Each element has two nodes, and each node has three degrees of freedom: lateral displacement, rotation, and longitudinal displacement. An element stiffness matrix is ​​established, with its elements determined by parameters such as the elastic modulus, moment of inertia, cross-sectional area, and shear correction factor. The stiffness matrices of all elements are then assembled into a global stiffness matrix according to the node connections.

[0039] Boundary conditions and loads are applied, specifically: elastic constraint boundary conditions are applied at the six clamping points to constrain the lateral displacement degrees of freedom. The constraint stiffness at the clamping points is determined based on the contact stiffness of the clamps, taking into account the elastic deformation of the contact interface and the local Hertzian contact effect. The constraint stiffness values ​​are obtained through prior contact mechanics simulations or experimental tests. The currently measured pressure values ​​at the six clamping points are applied as boundary constraint forces to the corresponding nodes.

[0040] The cutting force is applied to the element corresponding to the current tool position and is represented as a distributed load. The magnitude of the cutting force is estimated based on the cutting parameters: based on the depth of cut, feed rate, and tool geometry, an empirical cutting force model is used for calculation. The cutting force is decomposed into three components: tangential force, radial force, and axial force, which are applied to their respective degrees of freedom.

[0041] A modal superposition method is employed for order reduction, specifically: a fast solution is obtained by using modal superposition, avoiding the computational burden of directly solving high-dimensional equations. The eigenvalue problem of the system is solved, yielding the first 20 natural frequencies and mode shapes. Each mode shape is a displacement field mode, describing the deformation of the structure at that frequency. The displacement field of the structure is expressed as a linear combination of the mode shapes, with the combination coefficients called modal coordinates.

[0042] The original high-dimensional mechanical equations are projected onto modal space, transforming them into a system of low-dimensional algebraic equations concerning modal coordinates. Since only the first 20 modes are retained, the dimension of the equations is reduced from 150 to 20, significantly decreasing the computational load. After solving for the modal coordinates, the displacement and stress fields of the original space are reconstructed using the modal superposition formula.

[0043] The stress and deformation fields are calculated, specifically by determining the strain and stress at each location on the guide rail based on the obtained displacement field. For the beam model, the normal stress is the superposition of bending and axial stresses, while the shear stress is generated by shear deformation. Within each element, the stress is linearly distributed along the section height, with the maximum stress occurring on the outer surface of the section. The Von Mises equivalent stress distribution across the entire guide rail field is calculated. Von Mises stress comprehensively considers the combined effects of normal and shear stresses and is used to assess the yield risk of the material. The steps for calculating Von Mises equivalent stress are as follows: Obtain the values ​​of the normal stress components in three directions and the values ​​of the three shear stress components. Then calculate the sum of the squares of the three normal stress components, and then calculate the sum of the products of the pairwise products of the normal stress components. Subtract the sum of the products from the sum of the squares, calculate the sum of the squares of the three shear stress components, multiply the sum of the squares of the shear stress by 3, add the two results together, and finally take the square root of the sum to obtain the Von Mises equivalent stress.

[0044] For thin-walled beam structures, the main stress components are axial normal stress and shear stress, while other components are relatively small. The maximum Von Mises stress in the overall stress distribution is extracted as a representative indicator of the guide rail's stress state. Simultaneously, the maximum lateral displacement in the displacement field is extracted as a representative indicator of the guide rail's deformation state.

[0045] The fusion of model prediction and actual measurements involves comparing the predicted displacements of the digital twin model at eight displacement measurement points with the actual displacement sensor measurements. The prediction error is calculated as the ratio of the absolute value of the difference between the two values ​​to the measured value. When the prediction error is less than 10%, the model is considered accurate and reliable, and the predicted stress and deformation results are used directly. When the prediction error is greater than 10%, it indicates that there are deviations in the model parameters or boundary conditions, requiring model correction or data fusion.

[0046] A weighted fusion strategy is adopted, which calculates a weighted average of model predictions and measured values ​​to obtain a fused state estimate. The fusion weights are dynamically adjusted based on model confidence and measurement noise: when the model prediction error is small, the model weight is larger, making full use of the model's full-field prediction capability; when the prediction error is large, the measured data weight is larger, based on actual measurements. The fused state estimate combines the predictive power of the model with the realism of the measurements, improving the accuracy and robustness of state perception.

[0047] After the above processing, the stress state (including the maximum Von Mises stress and its location), deformation state (including the maximum displacement and its location), and estimated fusion displacement values ​​of each measuring point of the guide rail under the current clamping force and cutting force are obtained. This information comprehensively describes the real-time mechanical state of the guide rail and provides a basis for subsequent safety assessment.

[0048] Furthermore, a digital twin model based on a physical information neural network can be used to replace the aforementioned finite element model, thereby further improving the model's computational efficiency and adaptability. The physical information neural network embeds the mechanical control equations as part of the loss function into the neural network training process, enabling the network to not only learn data patterns but also follow physical laws. Specifically, a multi-layer feedforward neural network is constructed, with inputs including spatial coordinates, clamping force distribution, and cutting force parameters, and outputs displacement and stress fields. During the training phase, the loss function includes two parts: the fitting error between the predicted value and the measured data; and the residual after substituting the displacement field into the elasticity equations. By minimizing the comprehensive loss function, the network weights are trained to obtain a surrogate model that conforms to physical laws and fits the actual data. During the inference phase, only the forward propagation calculation of the neural network is required, which is hundreds of times faster than traditional finite element solutions. As measured data accumulates during the production process, incremental training can be continuously performed to update the network parameters, allowing the model to continuously adapt to the characteristics of the actual system, achieving self-learning and self-evolution. The purpose of using the physical information neural network method is to significantly improve computational speed while ensuring prediction accuracy, while endowing the model with online learning and adaptive capabilities to better cope with changes in material properties and uncertainties in operating conditions.

[0049] Step 4: Based on the analysis results of guide rail stress and deformation state and vibration sensing data, assess stress safety, deformation stability and vibration stability, and generate safety level; This step, based on the stress and deformation states predicted by the digital twin model and combined with the actual measured displacement and vibration data, adopts a multi-dimensional safety assessment criterion to obtain the safety level and adjustment recommendations for the current clamping state.

[0050] The stress safety assessment involves calculating the stress ratio based on the maximum Von Mises stress predicted by the digital twin model and the material's actual yield strength. The steps are as follows: Obtain the maximum Von Mises stress value predicted by the digital twin model, obtain the actual yield strength value of the material, and divide the maximum Von Mises stress by the actual yield strength to obtain the stress ratio.

[0051] The stress ratio reflects how close the current stress level of the guide rail is to the material's bearing capacity limit. A stress safety threshold of 0.8 is defined, with a 20% safety margin. When the stress ratio is less than or equal to the safety threshold, the stress is considered safe; the guide rail is within its elastic deformation range and no permanent plastic deformation will occur. When the stress ratio is greater than the safety threshold, an overload risk is identified; the stress is close to or exceeds the material's yield strength, posing a risk of plastic deformation and workpiece scrap. In this case, the clamping force needs to be reduced to lower the stress level.

[0052] Deformation stability assessment is conducted, specifically based on the maximum displacement predicted by the digital twin model and the machining accuracy requirements, to evaluate deformation stability. The machining accuracy requirement is that the straightness tolerance of the guide rail in the length direction should not exceed 0.1 mm. To ensure that real-time deformation during machining does not affect the final accuracy, a deformation threshold is defined as 50% of the tolerance, i.e., a deformation threshold of 0.05 mm. When the maximum displacement is less than or equal to the deformation threshold, the deformation is considered to be within an acceptable range, indicating good machining stability. When the maximum displacement exceeds the deformation threshold, the risk of excessive deformation is identified, indicating insufficient guide rail stiffness. Excessive deformation under cutting force will lead to dimensional deviations and decreased surface quality, requiring an increase in clamping force to improve workpiece stiffness.

[0053] Vibration stability assessment, specifically, involves calculating the vibration amplitude based on actual measured triaxial vibration acceleration, using the following steps: Obtain the vibration acceleration component values ​​in the x-axis direction, the y-axis direction, and the z-axis direction. Square each of the three components, add the three squared values ​​together, and finally take the square root of the sum to obtain the vibration acceleration amplitude.

[0054] Simultaneously, spectral analysis is performed on the vibration signal, and the power spectral density of the acceleration signal is calculated using Fast Fourier Transform. The main frequency components in the spectrum are identified to determine the presence of characteristic frequencies of cutting chatter (typically in the range of hundreds to thousands of hertz). A vibration amplitude threshold of 5g is defined; when the vibration amplitude is less than or equal to the threshold and there are no obvious chatter peaks in the spectrum, the vibration is considered stable, and the machining process is smooth.

[0055] When the vibration amplitude exceeds the vibration amplitude threshold or the spectrum shows chatter characteristics, it is determined that the vibration is too large. The workpiece will generate chatter during the cutting process, which will affect the surface quality of the machined surface and may cause tool damage. It is necessary to increase the clamping force to suppress the vibration.

[0056] A comprehensive safety level assessment is conducted, specifically based on the evaluation results of three dimensions: comprehensive stress safety, deformation stability, and vibration stability, to generate a comprehensive safety level for the clamping state. When the stress ratio is less than or equal to the stress safety threshold, the maximum displacement is less than or equal to the deformation threshold, and the vibration amplitude is less than or equal to the vibration amplitude threshold, and all three indicators meet the safety requirements, the overall safety level is determined to be "safe". The current clamping force configuration is reasonable and no adjustment is needed. Continue processing in the current state.

[0057] Among them, the stress safety threshold is dimensionless; the deformation threshold is in millimeters.

[0058] When any one of the three indicators is close to the threshold (deviation within 10% of the threshold) but has not yet exceeded it, the overall safety level is determined to be "warning". This indicates that the current clamping state is acceptable but there is a potential risk. Precautionary adjustments are recommended to avoid exceeding the safety range when the working conditions change.

[0059] If any one of the three indicators exceeds the threshold, the overall safety level is determined to be "dangerous". The clamping force must be adjusted immediately, otherwise it will lead to workpiece scrapping or processing accidents.

[0060] Based on the safety level assessment results, corresponding adjustment suggestions are generated: if the stress ratio is too high, it is recommended to reduce the clamping force to reduce stress; if the deformation or vibration is too large, it is recommended to increase the clamping force to improve stiffness and stability; if multiple indicators exceed the standard at the same time, a comprehensive balance needs to be made, prioritizing stress safety, while taking deformation and vibration control into account on this basis.

[0061] After the above processing, the overall safety level of the current clamping state ("safe", "warning" or "dangerous") and specific adjustment suggestions are obtained, providing clear goals and constraints for subsequent clamping force optimization decisions.

[0062] Step 5: Adjust the current clamping force distribution according to the safety level to obtain the clamping force configuration result; Based on the safety assessment results and adjustment suggestions, this step uses a multi-objective particle swarm optimization algorithm to obtain the optimal clamping force distribution scheme that balances stress minimization, deformation minimization, and vibration suppression while meeting safety constraints.

[0063] An optimization model is established, specifically defined as follows: the optimization variables are the clamping forces at six clamping points, denoted as F1, F2, F3, F4, F5, and F6, in Newtons. A multi-objective optimization function is established, containing three sub-objectives: stress minimization, deformation minimization, and vibration suppression.

[0064] The three sub-objectives of optimizing the objective function are specifically manifested as follows: The calculation steps for the first objective function are as follows: obtain the displacement values ​​of all nodes, find the maximum displacement value, and use this maximum displacement value as the objective to be minimized. The first objective function represents the goal of minimizing the maximum displacement; the clamping force vector represents the combination of clamping forces at the six clamping points; the displacement of the i-th node represents the displacement of the i-th node in the finite element model, in millimeters.

[0065] The calculation steps for the second objective function are as follows: obtain the numerical value of the processing stability index, and use this stability index as the objective to be maximized. The second objective function represents the goal of maximizing processing stability; the processing stability index represents the overall stability level of the system and is dimensionless.

[0066] Define a comprehensive stability index that comprehensively considers vibration suppression and stress margin. The calculation steps are as follows: Obtain the weighting coefficients for vibration suppression and stress margin. Then, obtain the current vibration amplitude and the normalized reference value of the vibration amplitude. Calculate the ratio of the vibration amplitude to the reference value. Subtract this ratio from 1 to obtain the vibration suppression term. Next, obtain the current maximum stress and the material yield strength. Calculate the ratio of the maximum stress to the yield strength. Subtract this ratio from 1 to obtain the stress margin term. Multiply the vibration suppression term by its weighting coefficient and the stress margin term by its weighting coefficient. Finally, add the two weighted terms together to obtain the comprehensive stability index.

[0067] The comprehensive stability index ranges from 0 to 1, with higher values ​​indicating better stability. The weighting coefficients for vibration suppression and stress margin both satisfy a normalization condition, with the sum of the two coefficients equal to 1. These coefficients are set according to the characteristics of the processing task; typically, the vibration suppression weighting coefficient is 0.6 and the stress margin weighting coefficient is 0.4. The current vibration amplitude is measured in grams (g). The normalized reference value for the vibration amplitude is twice the threshold value, used to normalize the vibration amplitude to the range of 0 to 1. The current maximum stress is measured in megapascals (MPa). The material yield strength is measured in MPa. The first item in the stability index represents the vibration suppression effect; the closer the value is to 1, the smaller the vibration. The second item represents the stress safety margin; the closer the value is to 1, the further the stress is from the yield limit.

[0068] The objective function expression is to maximize the overall stability index, which is equivalent to minimizing the negative stability index.

[0069] The calculation steps for the third objective function are as follows: obtain the power consumption value of each actuator, obtain the working time of each actuator, multiply the power consumption of each actuator by its corresponding working time to obtain the energy consumption of that actuator, add up the energy consumption of all actuators to obtain the total clamping energy consumption, and use this total energy consumption as the objective to be minimized. The third objective function represents the minimization objective of the total clamping energy consumption; the power consumption of the j-th actuator is in watts; the working time of the j-th actuator is in seconds; the total number of actuators represents the number of actuators in the system.

[0070] The third objective function minimizes the total clamping energy consumption, avoiding over-clamping. Clamping energy consumption is proportional to the sum of clamping forces, and the objective function expression is to minimize the sum of clamping forces at the six clamping points.

[0071] In this system, the clamping force at the i-th clamping point is expressed in Newtons; the summation symbol represents the accumulation of the clamping forces at all six clamping points; the stress minimization objective function is to minimize the maximum Von Mises stress, calculated by using a digital twin model to determine the stress distribution under a given clamping force configuration, and extracting the maximum stress value as the objective function value. The deformation minimization objective function is to minimize the maximum displacement, also calculated using a digital twin model to determine the displacement field, and extracting the maximum displacement value. The vibration suppression objective function is to minimize the amplitude of the vibration transfer function, which, based on structural dynamics theory, reflects the vibration amplification factor from external excitation to the response point, and is obtained through modal analysis.

[0072] The three single-objective functions are weighted and combined to form a comprehensive objective function. The weight coefficients are dynamically adjusted according to the current safety level: when stress exceeds the limit, the weight of stress minimization is increased; when deformation exceeds the limit, the weight of deformation minimization is increased; when vibration exceeds the limit, the weight of vibration suppression is increased.

[0073] The constraints are defined as follows: Upper and lower bounds for the clamping force are set, with the clamping force at each clamping point ranging from 200N to 2000N. The lower bound ensures sufficient clamping force to maintain workpiece fixation, while the upper bound prevents excessive clamping force from causing workpiece deformation or damage. Uniformity constraints for the clamping force distribution are also set, requiring the difference in clamping force between adjacent clamping points to not exceed 500N to avoid localized stress concentration. Safety constraints are further set, requiring the optimized clamping force configuration to meet the requirements of a stress ratio less than a safety threshold, a maximum displacement less than a deformation threshold, and a vibration amplitude less than a vibration threshold.

[0074] The particle swarm optimization (PSO) algorithm is implemented, specifically by using it to solve a multi-objective optimization problem. The particle swarm is initialized with 30 particles, each representing a six-dimensional clamping force configuration. Initial particle positions are randomly generated to ensure constraints are met. Velocity and position update rules are defined, including inertia, cognitive, and social terms. The inertia weight uses a linear decreasing strategy, decreasing from 0.9 to 0.4 to balance the algorithm's global and local search capabilities. Both the cognitive and social coefficients are set to 2.0 to encourage particles to move towards their individual and global optimum positions.

[0075] In each iteration, the objective function value is calculated for each particle using a digital twin model to evaluate its fitness. The individual optimal position and the global optimal position are updated. When a particle's position exceeds the constraints, a boundary handling strategy is used to pull it back into the feasible region. The convergence criterion is set as either a change in the global optimal value of less than 1% over 20 consecutive iterations or reaching the maximum number of iterations (200).

[0076] Choosing the optimal solution involves selecting the best solution from the particle swarm as the recommended clamping force configuration after the algorithm converges. For multi-objective optimization problems, the concept of Pareto optimality is adopted, selecting the solution that is not dominated by other solutions on any objective. If multiple Pareto optimal solutions exist, a trade-off is made based on the current safety level and adjustment requirements: if the current state is "dangerous," the solution that can quickly reduce risk is prioritized; if the current state is "warning," the solution with the best overall performance is selected; if the current state is "safe" but requires further optimization, the solution that can improve processing quality is selected.

[0077] The selected optimal solution is verified for feasibility by recalculating its corresponding stress, deformation, and vibration states using the digital twin model to confirm that all constraints are met. If the verification passes, the optimal clamping force configuration is output; if the verification fails, a suboptimal solution is selected for re-verification until a feasible optimal solution is found.

[0078] After the above processing, a six-point clamping force distribution scheme that meets safety constraints and has optimal performance is obtained, including the target clamping force value of each clamping point and the expected performance improvement effect, providing a clear control target for subsequent precise execution of clamping force.

[0079] Step 6: Calculate the clamping force adjustment amount based on the clamping force configuration result and convert it into a control command for adjustment to obtain the clamping force distribution result; process the guide rail based on the clamping force distribution result and perform quality inspection to obtain the processing quality inspection result; This step, based on the optimized target clamping force configuration, uses a precision clamping actuator driven by a servo motor. Through force closed-loop control, the clamping force is precisely adjusted to obtain the actual clamping force distribution after precise adjustment.

[0080] The clamping force adjustment is calculated by comparing the target clamping force output by the optimization algorithm with the current actual clamping force, and then calculating the required clamping force adjustment for each clamping point. The clamping force adjustment equals the target clamping force minus the current clamping force; a positive value indicates that the clamping force needs to be increased, and a negative value indicates that the clamping force needs to be decreased.

[0081] The magnitude of the adjustment is determined, and a minimum adjustment threshold of 50N is defined. When the absolute value of the adjustment is less than the minimum adjustment threshold, the current clamping force at that clamping point is considered sufficiently close to the target value, and no adjustment is needed to avoid wear on the actuator caused by frequent minor adjustments. When the absolute value of the adjustment is greater than or equal to the minimum adjustment threshold, the clamping force at that clamping point is adjusted.

[0082] The system generates control commands for the actuators, specifically by converting the target clamping force into control commands for the servo motors at the clamping points that require adjustment. Each clamping point is equipped with a servo motor-driven ball screw clamping actuator. The ball screw lead is 5mm, meaning that the screw moves axially by 5mm for every revolution of the motor. The mechanical transmission ratio of the clamping mechanism is 10:1, meaning that the clamping force change generated by a 1mm movement of the ball screw is 500N.

[0083] Based on the target clamping force increment, the required lead screw displacement increment is calculated by dividing the clamping force increment by 500mm. Then, the required motor rotation angle increment is calculated by dividing the displacement increment by 5 and multiplying by 360 degrees. This angle increment is then converted into the number of pulses from the motor encoder and input to the servo driver.

[0084] To ensure a smooth adjustment process, an S-curve speed curve is used to plan the motor's motion trajectory. The S-curve includes three stages: acceleration, constant speed, and deceleration, avoiding shocks and vibrations caused by sudden speed changes. The acceleration and jerk of the S-curve are continuous, further preventing vibrations caused by sudden speed changes. The time scale for speed planning is set to 30 milliseconds to ensure a fast and smooth adjustment process.

[0085] The force closed-loop control is specifically implemented as follows: during the clamping force adjustment process, force closed-loop control ensures that the actual clamping force accurately converges to the target value. The servo drive operates in torque control mode, and based on the deviation between the clamping force and the target clamping force fed back in real time by the pressure sensor, a PID control algorithm is used to calculate the control quantity and dynamically adjust the motor output torque.

[0086] The calculation steps of the PID control algorithm are as follows: The PID control process calculates the current clamping force deviation, which is the target clamping force minus the actual clamping force. It then calculates the proportional term (deviation multiplied by a proportional gain), the integral term (cumulative sum of historical deviations multiplied by an integral gain), and the derivative term (difference between the current deviation and the deviation at the previous moment divided by the sampling time and multiplied by a derivative gain). The proportional, integral, and derivative terms are then summed to obtain the PID control output, which serves as the motor torque command. The proportional gain is set to 0.8, the integral gain to 0.2, and the derivative gain to 0.1. The control cycle is 1 millisecond to ensure rapid response.

[0087] During the adjustment process, the clamping force deviation is monitored in real time. When the absolute value of the deviation is less than 2% of the target value and is maintained for 10 consecutive control cycles, it is determined that the clamping force at the clamping point has stabilized and converged to the target value, the adjustment is stopped, and the current state is locked.

[0088] The verification of the adjustment effect and feedback is as follows: After the clamping forces at all six clamping points have been adjusted and stabilized, sensor data is re-acquired to obtain the actual clamping force distribution, displacement data, and vibration data after the adjustment. A digital twin model is then invoked, inputting the adjusted clamping force configuration, to calculate the expected stress and deformation state. The model's predicted values ​​are compared with the actual sensor measurements to verify whether the adjustment effect has met expectations.

[0089] Check if the stress ratio has decreased below the safety threshold, if the maximum displacement has decreased below the stability threshold, and if the vibration amplitude has decreased below the stability threshold. If all three indicators meet the requirements, the adjustment is considered successful, the overall safety level is restored to "safe," and processing continues with the current clamping force configuration.

[0090] If the indicators still exceed the standard after adjustment, analyze the possible reasons: If there is a large deviation between the actual clamping force and the target value, it may be due to insufficient precision of the actuator or sensor calibration error, requiring recalibration or adjustment of control parameters; If the clamping force reaches the target value but the performance indicators still do not meet the requirements, it may be due to deviation in the model prediction of the optimization algorithm, requiring updating the model parameters or recalculating the optimization.

[0091] Record key data from the adjustment process, including the clamping force distribution before and after adjustment, changes in performance indicators, and adjustment time, as historical data for subsequent online learning and algorithm optimization.

[0092] After the above processing, a precisely adjusted six-point clamping force distribution is obtained. The deviation between the actual clamping force and the target value is less than 2%. The stress state and deformation state of the guide rail meet the safety and stability requirements. The system is restored to a safe working state and can continue high-quality processing operations.

[0093] Step 7: Update the quality prediction model based on the clamping force distribution results and machining quality inspection results, predict tool wear and calculate compensation, and obtain the clamping force control scheme; This step, based on historically accumulated adjustment data and machining quality inspection results, uses an incremental learning algorithm to continuously update the quality prediction model and optimize algorithm parameters, resulting in an updated adaptive control strategy and tool wear compensation suggestions.

[0094] Accumulating historical data and extracting features are specifically manifested in the following ways: During the production process, the system continuously records key data for each processing batch, including: identified material parameters (elastic modulus, yield strength), cutting parameters (depth of cut, feed rate, spindle speed), optimized clamping force configuration scheme, stress and deformation monitoring data during processing, and quality inspection results after processing (surface roughness, dimensional accuracy, straightness, flatness, etc.).

[0095] Feature vectors are extracted from the raw data and used as input to the machine learning model. Features include material property-related features (deviation rate of elastic modulus from nominal value, deviation rate of yield strength from nominal value, operating temperature), machining condition-related features (depth of cut, feed rate, estimated cutting force, cumulative tool usage time), and clamping force configuration-related features (average, standard deviation, maximum, minimum, and distribution non-uniformity of six-point clamping forces). Labels are machining quality indicators (surface roughness Ra value, dimensional accuracy deviation, straightness error, and flatness error).

[0096] Historical data is organized chronologically, with the most recent data given higher weights and earlier data having their weights diminish over time, reflecting the time-varying nature of the production system's characteristics.

[0097] The online quality prediction model is updated by establishing a correlation model from clamping force configuration and processing conditions to processing quality, and using an incremental support vector regression algorithm for online learning. Support vector regression finds the optimal hyperplane to fit the input-output relationship, exhibiting good generalization ability and the ability to model nonlinear relationships.

[0098] The prediction function for incremental support vector regression is: The calculation steps of the SVR prediction function are as follows: obtain the number of support vectors, and perform the following operations on each support vector: obtain the Lagrange multiplier corresponding to the support vector, calculate the difference of the Lagrange multipliers, obtain the kernel function value between the support vector and the input vector, multiply the difference of the Lagrange multipliers by the kernel function value, add the calculation results of all support vectors, and finally add the bias term to obtain the prediction output.

[0099] The specific implementation of the kernel function: The radial basis function (RBF) is used as the kernel function, and its calculation steps are as follows: Obtain the bandwidth parameter of the kernel function, obtain the input vector and support vectors, calculate the square of the difference between the input vector and support vectors in each dimension, add up the squared differences of all dimensions to obtain the square of the Euclidean distance, multiply the squared distance by the negative bandwidth parameter, and calculate the kernel function value through the exponential function.

[0100] The bandwidth parameter of the kernel function controls its width; a larger bandwidth parameter value results in a narrower kernel and higher model complexity. The squared Euclidean distance between the input vector and support vectors is also considered. The RBF kernel function maps nonlinear relationships in the input space to a high-dimensional feature space, making linearly inseparable problems in the original space linearly separable in the feature space, thus improving the model's nonlinear fitting ability. The kernel function is calculated by multiplying the sum of the squares of the differences between the input vector and support vectors in each dimension by a negative bandwidth parameter, and finally obtaining the kernel function value through an exponential function mapping.

[0101] In the initial stage, a basic model is trained using a small amount of initial data to obtain an initial set of support vectors and model parameters. After each batch of guide rails is processed, the feature vectors and quality labels of that batch are obtained as new samples. An incremental learning algorithm is used to update the model: it is determined whether the new sample has an improvement effect on the current model. If the prediction error of the new sample is large or it is located in the boundary region of the feature space, it is added to the support vector set and the model parameters are updated; if the new sample is within the good fit range of the current model, it does not need to be added to avoid the model size growing indefinitely.

[0102] Regularly evaluate the model's performance using recent batches of data as a validation set and calculate the model's prediction error. A sustained decrease in prediction error indicates improved model adaptability; an upward trend in prediction error indicates significant changes in the characteristics of the production system, triggering global retraining or parameter retuning of the model.

[0103] The adaptive adjustment of the optimization algorithm parameters is specifically manifested in the following way: based on the updated quality prediction model, the parameters of the multi-objective optimization algorithm are adjusted in reverse to make the subsequent clamping force optimization decision more in line with the actual production requirements.

[0104] Analyze the correlation patterns revealed by the quality prediction model to identify the factors that have the greatest impact on processing quality. If the model shows that surface roughness is highly sensitive to vibration suppression, increase the weight of the vibration suppression term in the stability index weighting. If the model shows that dimensional accuracy is highly sensitive to maximum deformation, increase the weight of deformation minimization in the objective function.

[0105] Adjust the constraint threshold based on the statistical distribution of actual processing quality. If the stress ratio remains below 0.6 in actual production, it indicates that the stress constraint is too conservative, and the stress safety threshold can be appropriately relaxed to improve material utilization. If situations frequently occur close to the threshold in actual production, it indicates that the threshold setting is too aggressive, and the threshold should be tightened to leave more safety margin.

[0106] Adjust the parameters of the particle swarm optimization algorithm, such as the number of particles, the inertia weight decay coefficient, and the cognitive and social coefficients. If the algorithm converges slowly, increase the initial value of the inertia weight or increase the number of particles to enhance the search capability; if the algorithm is prone to getting trapped in local optima, increase the cognitive coefficient to encourage particles to explore new regions.

[0107] Estimating tool wear and compensation involves tracking the cumulative machining time and cutting force trends to estimate the degree of tool wear. As tools are used, the cutting edge gradually wears down, the cutting force gradually increases, and the surface finish deteriorates. By monitoring the long-term trend of cutting force changes, the increase in cutting force caused by tool wear can be identified.

[0108] An empirical model of tool wear and cutting force increment was established, based on historical data statistics. For a specific tool type and machining material, the cutting force increment and tool wear degree show an approximately linear relationship. The expected cutting force increment is obtained by consulting the empirical model based on the cumulative tool usage time.

[0109] The incremental cutting force is input into the digital twin model and optimization algorithm to pre-calculate the required clamping force increment to cope with tool wear. When the tool wear reaches a preset threshold (e.g., a 20% increase in cutting force), the system automatically issues a tool replacement prompt to avoid a decline in machining quality due to excessive tool wear.

[0110] The assessment and proactive exploration of operating condition space coverage specifically involves evaluating whether the system's accumulated operational experience covers the possible operating conditions. The dimensions of the operating condition space are defined, including the range of material property variations, the range of cutting parameter variations, and the range of ambient temperature variations. The distribution density of historical data across each dimension is calculated to identify sparse data areas.

[0111] For operating conditions with sparse data, the system proactively suggests conducting exploratory experiments, trying new parameter combinations while ensuring safety, and accumulating more comprehensive operational experience. The exploration strategy employs an uncertainty-based active learning method, prioritizing operating conditions with higher model prediction uncertainty for experimentation to maximize information acquisition efficiency.

[0112] Through proactive exploration, the system can continuously expand its adaptability, improve its generalization performance to unknown operating conditions, and reduce the risk of control failure caused by changes in operating conditions.

[0113] After the above processing, a continuously updated quality prediction model, adaptively adjusted optimization algorithm parameters, tool wear compensation strategy, and expanded working condition adaptability are obtained, enabling the entire intelligent control system to have the ability to learn online and evolve on its own, continuously adapt to changes in the production environment, and maintain long-term stable control performance.

[0114] By organically combining the above seven steps, this invention constructs a complete closed-loop system for intelligent control of clamping force in sunroof guide rail fixtures. Starting with multi-source sensing, the system achieves intelligent and precise control of clamping force through material property identification, digital twin modeling, safety assessment, multi-objective optimization, precision execution, and online learning. The system possesses advanced functions such as real-time response, adaptive learning, and predictive maintenance, improving the quality stability and production efficiency of thin-walled guide rail processing, and providing important technical support for intelligent manufacturing.

[0115] A smart control system for the clamping force of a sunroof guide rail clamp, such as Figure 2As shown, a method for intelligently controlling the clamping force of a sunroof guide rail fixture, as described above, includes: The sensor data acquisition module is used to acquire multi-dimensional sensor data of the sunroof guide rail fixture and preprocess it to obtain a multi-source synchronous sensor dataset, including displacement sensor data, temperature sensor data, current clamping force distribution and vibration sensor data. The material property acquisition module is used to obtain force-displacement data pairs from displacement sensing data using a micro-loading test method and identify the compliance coefficient of the guide rail; the elastic modulus is calculated from the compliance coefficient and temperature correction is performed by combining it with temperature sensing data to obtain the actual material property parameters. The state analysis and calculation module is used to calculate the theoretical stress distribution and displacement field based on the actual material property parameters and the current clamping force distribution, and to fuse and correct them with the sensor measured data to obtain the stress and deformation state analysis results of the guide rail. The safety assessment module is used to assess stress safety, deformation stability, and vibration stability based on the analysis results of guide rail stress and deformation state and vibration sensing data, and to generate a safety level. The clamping force configuration module is used to adjust the current clamping force distribution according to the safety level and obtain the clamping force configuration result. The execution adjustment control module is used to calculate the clamping force adjustment amount based on the clamping force configuration result and convert it into control commands for adjustment, thereby obtaining the clamping force distribution result; based on the clamping force distribution result, the guide rail is machined and quality inspection is performed to obtain the machining quality inspection result; The adaptive learning module is used to update the quality prediction model based on the clamping force distribution results and machining quality inspection results, predict tool wear and calculate compensation, and obtain a clamping force control scheme.

[0116] In one embodiment of the present invention, a specific example is provided: To verify the effectiveness of the method of this invention, a 30-day field test was conducted on the sunroof guide rail production line of an automotive parts manufacturing company. During the test, a complete intelligent control system was deployed, including core modules such as a multi-dimensional sensor network, a digital twin modeling platform, a multi-objective optimization algorithm, and precision actuators. The test area covered approximately 500 square meters of the entire production workshop, involving the coordinated operation of six production lines. The test object was 6063 aluminum alloy extruded guide rails, with a total of 1200 guide rail samples processed, covering materials from five different suppliers. The ambient temperature range was 18–28 degrees Celsius, fully verifying the system's adaptability and stability in the actual production environment.

[0117] The table below shows examples of multi-source sensor data acquired by the system under typical operating conditions: Table 1: Measured pressure values ​​at six clamping points under typical working conditions; Table 2: Measured values ​​of three displacement measuring points, vibration shaft and temperature under typical working conditions; Tables 1 and 2 together present the real-time sensor data acquired by the system, including the pressure distribution at the six clamping points, micro-displacement changes at key locations, triaxial vibration characteristics, and ambient temperature information. The data shows that the pressure at each clamping point remains within the range of 1800–1900 N, displacement changes are controlled at the micrometer level, vibration amplitude is maintained at a low level, and the temperature is stable at 23.5–23.6 degrees Celsius, demonstrating the system's good control accuracy and stability.

[0118] Through the detailed description of the above specific embodiments, the intelligent control method for clamping force of the sunroof guide rail fixture proposed in this invention effectively solves the problem of clamping force control in the thin-walled guide rail processing through a complete closed loop of multi-dimensional perception, digital twin modeling, adaptive optimization, precision execution and online learning. It realizes intelligent manufacturing with high precision, high quality and high stability, and has important practical application value.

[0119] The embodiments of the present invention have been described above. However, the embodiments are not limited to the specific implementation methods described above. The specific implementation methods described above are merely illustrative and not restrictive. Those skilled in the art can make more equivalent embodiments under the guidance of the present embodiments, and all of them are within the protection scope of the present embodiments.

Claims

1. A method for intelligently controlling the clamping force of a sunroof guide rail clamp, characterized in that, include: The multi-dimensional sensing data of the sunroof guide rail fixture is acquired and preprocessed to obtain a multi-source synchronous sensing dataset, including displacement sensing data, temperature sensing data, current clamping force distribution and vibration sensing data. The force-displacement data pairs were obtained by using a micro-loading test method on the displacement sensing data, and the compliance coefficient of the guide rail was identified. The elastic modulus is calculated by back-calculating the flexibility coefficient and temperature correction is performed by combining temperature sensing data to obtain the actual material property parameters. Based on actual material properties and current clamping force distribution, theoretical stress distribution and displacement field are calculated and fused with sensor measured data for correction to obtain the stress and deformation state analysis results of the guide rail. Based on the analysis results of guide rail stress and deformation state and vibration sensing data, stress safety, deformation stability and vibration stability are evaluated, and a safety level is generated. Adjust the current clamping force distribution according to the safety level to obtain the clamping force configuration result; The clamping force adjustment amount is calculated based on the clamping force configuration result and converted into control commands for adjustment, thus obtaining the clamping force distribution result; the guide rail is machined based on the clamping force distribution result and quality inspection is performed to obtain the machining quality inspection result; The quality prediction model is updated based on the clamping force distribution results and machining quality inspection results. Tool wear is predicted and compensation is calculated to obtain a clamping force control scheme.

2. The intelligent control method for clamping force of a sunroof guide rail clamp according to claim 1, characterized in that, The acquisition of the multi-dimensional sensor data includes: Obtain the physical structure information of the sunroof guide rail clamping system, and install piezoresistive pressure sensors at each clamping point based on the structural characteristics of the six-point clamping fixture to measure the normal force applied at the clamping point in real time. Laser displacement sensors are deployed at key locations on the guide rail to detect minute deformations of the guide rail; A triaxial accelerometer is installed at the center of the fixture body to monitor the vibration characteristics during the machining process; Infrared temperature sensors are placed above the working area to monitor the temperature distribution on the guide rail surface and obtain multi-dimensional physical quantity data.

3. The intelligent control method for clamping force of a sunroof guide rail clamp according to claim 1, characterized in that, The acquisition of the actual material property parameters includes: The preprocessed multi-source synchronous sensing dataset is acquired, and the clamping system is controlled to apply a preset small-amplitude clamping force sequence to the preprocessed multi-source synchronous sensing dataset. The deformation response of the displacement measuring point is collected synchronously to obtain force-displacement data pairs. Based on force and displacement data, a linear relationship model is established using the least squares method. The flexibility coefficient of the linear relationship model is then identified. The identified flexibility coefficient is then back-calculated using the guide rail geometric parameters to obtain the actual elastic modulus. The actual elastic modulus and yield strength are corrected based on the average temperature of the guide rail measured by the temperature sensor.

4. The intelligent control method for clamping force of a sunroof guide rail clamp according to claim 1, characterized in that, The acquisition of the stress and deformation state analysis results of the guide rail includes: The actual material properties and current clamping force distribution are obtained, and the guide rail is discretized using the Timoshenko beam model, which is then discretized into beam elements. Elastic constraint boundary conditions are applied to the beam elements at the six clamping points, and cutting loads are applied to the elements corresponding to the current tool position. The modal superposition method is used to solve the beam element after applying load with reduced order, and the theoretical stress distribution and displacement field are calculated. The theoretical calculation results are fused and corrected with the sensor's measured data.

5. The intelligent control method for clamping force of a sunroof guide rail clamp according to claim 1, characterized in that, The generation of the security level includes: Obtain the stress and deformation state analysis results of the guide rail and the measured vibration data. Calculate the stress ratio based on the maximum Von Mises stress and the actual yield strength of the material in the analysis results, and assess the stress safety. The deformation stability is evaluated by comparing the maximum displacement in the analysis results with the preset deformation threshold. Based on the analysis of measured vibration data, vibration amplitude and spectral characteristics are analyzed to assess vibration stability. The safety level is generated by combining the assessment results of stress safety, deformation stability, and vibration stability.

6. The intelligent control method for clamping force of a sunroof guide rail clamp according to claim 1, characterized in that, The solution for the clamping force configuration includes: Obtain safety assessment results, and when the safety level is warning or danger, establish a multi-objective optimization model based on the assessment results with the objectives of minimizing stress, minimizing deformation, and suppressing vibration. The particle swarm optimization algorithm is used to solve the optimization model, and the optimization efficiency is improved by using a hot start strategy and adaptive inertia weight. The optimal solutions are sorted by multi-objectives based on the Pareto dominance relationship, and the optimal solution is selected as the clamping force configuration scheme.

7. The intelligent control method for clamping force of a sunroof guide rail clamp according to claim 1, characterized in that, The results of obtaining the clamping force distribution include: Obtain the optimal clamping force configuration result, calculate the clamping force adjustment amount at each clamping point based on the configuration result, and convert the adjustment amount into servo motor control commands; An S-shaped speed curve is used to smoothly adjust and control the servo motor, and a force closed-loop PID control is used to precisely adjust the clamping force. The effect of the adjusted clamping force was verified, and the data of the adjustment process and the final clamping force distribution were recorded.

8. The intelligent control method for clamping force of a sunroof guide rail clamp according to claim 1, characterized in that, The clamping force control scheme is obtained by: Obtain historical adjustment data and processing quality results, and update the parameters of the quality prediction model based on the historical data using an incremental learning algorithm; The quality prediction model was rebuilt and trained using support vector regression. The updated model is evaluated for performance, and the algorithm parameters of the control strategy are optimized based on the evaluation results. Retrain the model periodically to maintain prediction accuracy.

9. The intelligent control method for clamping force of a sunroof guide rail clamp according to claim 1, characterized in that, The process of predicting tool wear and calculating compensation includes: Acquire cutting force variation data during the cutting process, and monitor the trend of the variation data to identify tool wear status; An empirical model of tool wear and cutting force increment is established based on tool wear status; Predict the incremental cutting force based on the empirical model according to the cumulative tool usage time; The clamping force compensation required to cope with wear is calculated based on the prediction results, and a tool replacement prompt is provided when the wear reaches a preset threshold.

10. A smart control system for the clamping force of a sunroof guide rail clamp, characterized in that, A method for intelligently controlling the clamping force of a sunroof guide rail clamp as described in any one of claims 1-9 includes: The sensor data acquisition module is used to acquire multi-dimensional sensor data of the sunroof guide rail fixture and preprocess it to obtain a multi-source synchronous sensor dataset, including displacement sensor data, temperature sensor data, current clamping force distribution and vibration sensor data. The material property acquisition module is used to obtain force-displacement data pairs from displacement sensing data using a micro-loading test method and identify the compliance coefficient of the guide rail; the elastic modulus is calculated from the compliance coefficient and temperature correction is performed by combining it with temperature sensing data to obtain the actual material property parameters. The state analysis and calculation module is used to calculate the theoretical stress distribution and displacement field based on the actual material property parameters and the current clamping force distribution, and to fuse and correct them with the sensor measured data to obtain the stress and deformation state analysis results of the guide rail. The safety assessment module is used to assess stress safety, deformation stability, and vibration stability based on the analysis results of guide rail stress and deformation state and vibration sensing data, and to generate a safety level. The clamping force configuration module is used to adjust the current clamping force distribution according to the safety level and obtain the clamping force configuration result. The execution adjustment control module is used to calculate the clamping force adjustment amount based on the clamping force configuration result and convert it into control commands for adjustment, thereby obtaining the clamping force distribution result; based on the clamping force distribution result, the guide rail is machined and quality inspection is performed to obtain the machining quality inspection result; The adaptive learning module is used to update the quality prediction model based on the clamping force distribution results and machining quality inspection results, predict tool wear and calculate compensation, and obtain a clamping force control scheme.

Images