A glass cover plate CNC adaptive feeding machining control method and system

Abstract

This invention relates to the field of CNC machining control technology, and discloses a CNC adaptive feed machining control method and system for glass cover plates. The method includes: acquiring the unloaded spindle mechanical output power and feedforward kinematic commands of the servo drive during the non-cutting idle stroke; calculating the equivalent friction coefficient and dynamic equivalent moment of inertia to establish a dynamic benchmark model; acquiring the cutting spindle mechanical output power during the actual cutting segment; calling the benchmark model to subtract non-cutting dissipation power to extract the pure cutting power component; reconstructing the actual cutting resistance vector angle from the pure cutting power component; comparing it with the theoretical trajectory tangential to obtain the normalized phase deviation; determining the tool micro-deflection polarity factor; and generating feed rate and spindle speed adjustment parameters based on the deviation and polarity factor to execute closed-loop control. This invention eliminates the interference of machine tool transmission losses on load detection, senses and corrects the tool force deflection state, avoids overcutting or breakage defects in the workpiece, and improves machining stability.

Description

Technical Field

[0001] This invention relates to the field of CNC machining control technology, specifically to a CNC adaptive feed machining control method and system for glass cover plates. Background Technology

[0002] In the manufacturing process of electronic products, glass covers are hard and brittle materials, and are quite sensitive to the stress on the cutting tool during CNC machining. If the machining stress is not properly controlled, defects such as edge chipping and overcutting can easily occur. Currently, CNC machine tools typically assess the machining load by monitoring the current or total power of the feed axis servo drives and passively adjust the feed speed according to the load.

[0003] In actual machining processes, the total load data output by the servo motor includes not only the actual cutting load generated by the tool removing material, but also the frictional losses of mechanical transmission components such as machine tool guideways and lead screws, as well as the inertial forces generated by acceleration and deceleration. Existing control methods fail to isolate these mechanical dissipation factors, resulting in significant deviations in the load data extracted by the system, which cannot accurately reflect the true force state of the tool during cutting. Due to the lack of pure cutting work data, the control system cannot perceive the microscopic deflection of the tool during machining.

[0004] Meanwhile, existing constant load control logic relies solely on scalar data such as total power for judgment, lacking the ability to extract vectors of cutting resistance direction. When machining complex contours, the control system cannot determine the physical trend of tool force deflection, nor can it identify whether the tool is squeezing inward or retreating outward. When abnormal load fluctuations occur, the system often can only adopt a simple overall machine speed reduction strategy, unable to implement independent interpolation axis intervention and coordinated spindle speed adjustment for specific overcut or undercut states. This single adjustment method is insufficient to maintain machining force balance when the tool undergoes micro-deflection, and cannot effectively avoid machining defects such as edge breakage or dimensional distortion of the glass cover. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a CNC adaptive feed machining control method and system for glass cover plates, which solves the problem that existing CNC machining control methods are unable to effectively separate the actual cutting load from mechanical transmission losses and cannot accurately sense and correct the microscopic force deflection state of the tool.

[0006] To achieve the above objectives, the present invention provides the following technical solution: The first aspect of this invention provides a CNC adaptive feed machining control method for glass cover plates, comprising the following steps: The mechanical output power and feedforward kinematic commands of the X-axis and Y-axis servo drives at the unloaded shaft ends are obtained during the non-cutting idle stroke. The dynamic equivalent friction coefficient and dynamic equivalent moment of inertia are calculated to establish a dynamic benchmark model. In the formal cutting section, the mechanical output power of the X-axis and Y-axis servo drives in the cutting state is obtained. The non-cutting dissipation power is calculated by calling the dynamic benchmark model in combination with the feedforward kinematic command. The pure cutting power component is extracted from the mechanical output power of the cutting state shaft end by subtracting the non-cutting dissipation power. The pure cutting power component is converted into the cutting resistance component, the actual cutting resistance vector angle is reconstructed, the normalized phase deviation is obtained by comparing the actual cutting resistance vector angle with the theoretical resistance reference angle, the absolute phase deviation and the sliding window change rate of the absolute phase deviation are obtained from the normalized phase deviation, and the micro-deflection polarity factor is determined based on the workpiece solid side normal. Real-time feed rate adjustment parameters are generated based on normalized phase deviation, absolute phase deviation, sliding window change rate, and micro-deflection polarity factor. Real-time spindle speed adjustment parameters are also generated based on the cutting load status. These parameters are fed back to the trajectory interpolator and spindle frequency converter to perform closed-loop control.

[0007] This invention establishes a mechanical transmission loss benchmark during the non-cutting idle stage, calculates and deducts the non-cutting dissipation power during the formal cutting stage, and extracts the pure cutting power component that reflects the actual machining load. Then, the force vector angle at the tool tip is calculated, and the micro-deflection polarity factor of the tool is determined by comparing the phase deviation between the actual force direction and the tangential direction of the theoretical trajectory. Based on the tool deflection state and evolution trend, the feed and spindle motion are adjusted synchronously. This scheme decouples machine tool friction and inertial load, and directly uses the deflection characteristics caused by pure cutting resistance to dynamically correct machining parameters.

[0008] Furthermore, the steps for establishing the dynamic benchmark model specifically include: determining whether the feed axis is in an idle state based on the instruction code of the CNC machining program and the axial coordinate of the tool spindle; when it is in an idle state and the absolute value of the feed axis instruction running speed exceeds the effective speed threshold, opening a preset dynamic identification time window; Within a time window, feedforward kinematic command parameters and the mechanical output power of the servo drive at the unloaded shaft end are extracted synchronously to construct a discrete sampling data set. The least squares regression algorithm is used to solve the sample set to obtain the dynamic equivalent friction coefficient and dynamic equivalent moment of inertia, so as to filter out the influence of the zero-speed dead zone and obtain the inherent dynamic parameters of the feed axis.

[0009] Furthermore, the steps for extracting the pure cutting power component specifically include: multiplying the product of the dynamic equivalent friction coefficient and the absolute value of the feedforward command running speed, and adding the product of the dynamic equivalent motion inertia, the feedforward command running acceleration, and the signed feedforward command running speed to obtain the non-cutting dissipation power of the X-axis and Y-axis respectively. The calculation result is obtained by subtracting the corresponding non-cutting dissipation power from the mechanical output power of the shaft end in the cutting state; when the calculation result is negative, the lower limit threshold is set to zero for amplitude limiting, and the calculation result after amplitude limiting is taken as the pure cutting power component.

[0010] Furthermore, after the step of extracting the pure cutting power component, the method further includes: multiplying the spindle speed by the effective number of cutting edges of the tool or the equivalent periodic excitation parameter and dividing by 60 to calculate the periodic excitation reference frequency, and combining it with a preset ratio to obtain the cutoff frequency of the first-order low-pass filter; and calculating the first-order filter coefficient by combining it with the inherent interpolation period of the trajectory interpolator. The pure cutting power component is subjected to first-order discrete low-pass filtering using first-order filtering coefficients. The filtered pure cutting power component is then used to reconstruct the actual cutting resistance vector angle and filter out high-frequency oscillations caused by intermittent cutting.

[0011] Furthermore, the steps of reconstructing the actual cutting resistance vector angle and comparing the actual cutting resistance vector angle with the theoretical trajectory tangential vector angle to obtain the normalized phase deviation specifically include: when the absolute value of the feed axis command running speed is greater than or equal to the micro-speed judgment threshold, the filtered pure cutting power component is divided by the corresponding feedforward command running speed, and the cutting resistance component is obtained by combining the direction correction. When the absolute value of the feed axis command running speed is less than the micro-speed judgment threshold, the cutting disturbance torque is obtained by subtracting the friction torque and inertial torque from the equivalent disturbance torque data of the servo driver. The cutting resistance component is obtained by torque conversion to prevent division by zero overflow. The four-quadrant arctangent function is called to synthesize the Y-axis and X-axis cutting resistance components to obtain the actual cutting resistance vector angle. The theoretical trajectory tangential vector angle is added to the constant of pi to obtain the theoretical resistance reference angle. The original phase difference is obtained by subtracting the actual cutting resistance vector angle from the theoretical resistance reference angle. Boundary verification and translation processing are performed on the original phase difference value, and the normalized phase deviation is output within the specified range.

[0012] Furthermore, the steps for determining the micro-deflection polarity factor specifically include: reading the tool radius compensation state command to determine the normal direction of the workpiece solid side; determining the two-dimensional rotation direction of the actual cutting resistance vector based on the normalized phase deviation; and projecting the deflection component of the actual cutting resistance vector relative to the theoretical resistance reference direction onto the normal direction of the workpiece solid side. If the projection result points into the interior of the workpiece, the micro-deflection polarity factor is assigned a value of 1; if the projection result deviates from the interior of the workpiece, the micro-deflection polarity factor is assigned a value of -1; if the projection result is lower than the preset polarity recognition threshold, the micro-deflection polarity factor is assigned a value of 0, thereby characterizing the physical trend of overcutting or undercutting.

[0013] Furthermore, the steps for generating real-time feed rate adjustment parameters specifically include: performing an absolute value operation on the normalized phase deviation to obtain the absolute phase deviation, and extracting the sliding window change rate of the absolute phase deviation using a data buffer window; When the absolute phase deviation is greater than the absolute phase deviation tolerance threshold, or the sliding window change rate is greater than the sliding window change rate tolerance threshold, and the micro-deflection polarity factor is assigned a value of 1, it is determined to be in an overcut state; calculate the first difference between the absolute phase deviation of the current interpolation period and the absolute phase deviation tolerance threshold, and the second difference between the sliding window change rate and the sliding window change rate tolerance threshold, and take the value of the difference that is less than zero as zero. The feed rate attenuation coefficient is obtained by multiplying the first difference by the deviation penalty weight and adding the second difference by the rate of change penalty weight. The intermediate feed rate is obtained by subtracting the feed rate attenuation coefficient from the base feed rate command value. When the intermediate feed rate is lower than the minimum safe feed rate threshold, the real-time feed rate is assigned to the minimum safe feed rate threshold. Otherwise, the intermediate feed rate is used as the real-time feed rate. The real-time feed rate is sent to the trajectory interpolator to perform dual-axis cooperative deceleration to suppress overcutting.

[0014] Furthermore, the step of generating real-time feed rate adjustment parameters also includes: when the absolute phase deviation is greater than the absolute phase deviation tolerance threshold, or the sliding window change rate is greater than the sliding window change rate tolerance threshold, and the micro deflection polarity factor is assigned a value of -1, it is determined to be in an undercut state. Extract the nominal feed increments of the X-axis and Y-axis for the current interpolation cycle. The axis with the larger absolute value is identified as the motion-dominant axis, and the axis with the smaller absolute value is identified as the driven-maintaining axis. When the absolute values ​​are equal, the X-axis is designated as the motion-dominant axis. Multiply the nominal feed increment of the motion-dominant axis by the attenuation coefficient to obtain the actual interpolation increment. The actual interpolation increment of the driven-maintaining axis is locked as the original nominal feed increment. The trajectory offset of a single interpolation cycle and the cumulative offset of a continuous slow-release cycle are limited. An updated interpolation increment command is issued to perform a one-way retreat to relieve the squeezing stress. After determining to resume synchronous interpolation, a reverse compensation increment is generated to return the nominal processing trajectory based on the cumulative trajectory offset.

[0015] Furthermore, the steps for generating real-time spindle speed ratio adjustment parameters specifically include: real-time reading of the spindle load rate, actual spindle speed, and the changing trend of the filtered pure cutting power component; when it is determined that the spindle is in an overcut state, reducing the real-time feed ratio and maintaining the spindle speed ratio at the reference state; When it is determined that the spindle is in an undercut state and the spindle load rate does not exceed the preset safe load threshold, a real-time spindle speed ratio adjustment parameter is generated, and the spindle speed ratio is adjusted between the preset upper limit ratio and the preset lower limit ratio.

[0016] A second aspect of this invention provides a CNC adaptive feed machining control system for glass covers, comprising a CNC main control unit, a real-time communication bus, and drive execution nodes, implementing the aforementioned machining control method; the CNC main control unit is equipped with: The parameter identification module acquires the mechanical output power and feedforward kinematic commands of the unloaded shaft ends of the X-axis and Y-axis servo drives during the non-cutting idle stroke, calculates the dynamic equivalent friction coefficient and dynamic equivalent moment of inertia, and establishes a dynamic benchmark model. The power stripping module obtains the mechanical output power of the shaft end in the cutting state during the formal cutting section, calls the dynamic benchmark model to calculate the non-cutting dissipation power, and extracts the pure cutting power component by subtracting the non-cutting dissipation power from the mechanical output power of the shaft end in the cutting state. The deflection decoupling module calculates and reconstructs the actual cutting resistance vector angle from the pure cutting power component, compares it with the theoretical trajectory tangential vector angle to obtain the normalized phase deviation, calculates the sliding window change rate and determines the micro deflection polarity factor. The compensation control module generates real-time feed rate adjustment parameters based on normalized phase deviation, sliding window change rate, and micro-deflection polarity factor, and generates real-time spindle speed adjustment parameters in conjunction with cutting load status.

[0017] This invention provides a CNC adaptive feed machining control method and system for glass cover plates. It has the following beneficial effects: 1. This invention establishes a dynamic benchmark by extracting the unloaded mechanical output power of the shaft end during the non-cutting idle travel phase and calculating the friction coefficient and dynamic equivalent moment of inertia. During the cutting phase, the non-cutting dissipated power is subtracted in real time from the mechanical output power of the shaft end in the cutting state to extract the pure cutting power component. This method effectively eliminates the interference of the machine tool transmission components' own friction and acceleration / deceleration inertia on load detection, and can accurately reflect the true stress state of the tool during actual machining.

[0018] 2. This invention reconstructs the actual cutting resistance vector angle using the extracted pure cutting power components and compares it with the theoretical trajectory tangential vector angle to generate a normalized phase deviation. This deviation is then combined with the workpiece normal to determine the microscopic deflection polarity factor. This processing logic converts cutting power data into geometric phase characteristics that characterize the direction of tool force deflection, clarifying the physical state of the tool tending towards overcutting or undercutting, and providing a quantitative basis for the adaptive adjustment of machining parameters.

[0019] 3. This invention performs closed-loop intervention based on the obtained phase deviation and deflection polarity for different cutting states. When overcutting is determined, it suppresses cutting deformation by reducing the speed through dual-axis coordinated action. When undercutting is determined, it alleviates compressive stress by changing the interpolation axis displacement ratio to perform unidirectional retraction. This adaptive adjustment mechanism can promptly correct the feed action in the early stages of tool micro-deflection, avoiding overcutting or breakage defects at the workpiece edge, and improving the stability of the machining process and product yield. Attached Figure Description

[0020] Figure 1 This is a flowchart of the method of the present invention; Figure 2 This is the logic diagram for the differential extraction of pure cutting power components in this invention; Figure 3 This is the logic diagram for the extraction of processing contour topology and mapping of microscopic deflection polarity factors in this invention; Figure 4 This is the logic diagram of the overcut state trigger determination and dynamic deceleration collaborative compensation of the present invention; Figure 5 This is a logic diagram of the undercut state trigger determination, single-axis slow release, and spindle coordinated adjustment of the present invention; Figure 6 This is a time-domain comparison curve of the pure cutting power component of the present invention; Figure 7 This is a graph showing the distribution of absolute phase deviation tracking data according to the present invention. Figure 8 This is a bar chart showing the statistical results of contour error and edge chipping defects in this invention. Detailed Implementation

[0021] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] This invention provides a CNC adaptive feed machining control system for glass cover plates, comprising a CNC main control unit, a real-time communication bus, and drive execution nodes. The CNC main control unit is internally configured with a trajectory interpolator. The drive execution nodes include an X-axis servo driver, a Y-axis servo driver, a Z-axis servo driver, and a spindle frequency converter driver. The CNC main control unit reads the tool spindle axial coordinates through position feedback data from the Z-axis servo driver or the machine tool coordinate register. The tool spindle axial coordinates are used to determine the idle state, cutting state, and to identify the interruption conditions of the time window. The CNC main control unit includes a parameter identification module, a power stripping module, a deflection decoupling module, and a compensation control module.

[0023] See attached document Figure 1 This invention provides a CNC adaptive feed machining control method for glass cover plates, comprising the following steps: S10, the parameter identification module reads the mechanical output power and feedforward kinematic commands of the unloaded shaft end of the X-axis servo drive and Y-axis servo drive during the non-cutting idle stroke, calculates the dynamic equivalent friction coefficient and dynamic equivalent moment of inertia, and establishes a dynamic benchmark model; S20, the power stripping module obtains the cutting state shaft end mechanical output power of the X-axis servo drive and Y-axis servo drive in the formal cutting section, and combines the command speed and acceleration data to call the dynamic benchmark model to calculate the non-cutting dissipation power. The cutting state shaft end mechanical output power is subtracted from the non-cutting dissipation power to extract the pure cutting power component. S30, the deflection decoupling module converts the pure cutting power component into the cutting resistance component, reconstructs the actual cutting resistance vector angle, compares the actual cutting resistance vector angle with the theoretical resistance reference angle to obtain the normalized phase deviation, obtains the absolute phase deviation and the sliding window change rate of the absolute phase deviation from the normalized phase deviation, and determines the micro deflection polarity factor based on the workpiece solid side normal. S40, the compensation control module generates real-time feed rate adjustment parameters based on normalized phase deviation, absolute phase deviation, sliding window change rate and micro deflection polarity factor, and generates real-time spindle speed adjustment parameters by combining spindle load rate, actual spindle speed and the change trend of filtered pure cutting power component. The real-time feed rate adjustment parameters are fed back to the trajectory interpolator, and the real-time spindle speed adjustment parameters are fed back to the spindle frequency converter to perform closed-loop intervention.

[0024] When the system is first started and before the identification of valid dynamic parameters is completed, the CNC main control unit calls the machine tool's factory calibration parameters or the dynamic reference parameters saved in the last valid machining task as initial values. If insufficient sampling points, incomplete rank of the observation matrix, loss of bus data, or abnormal servo power data occur during the parameter identification process, the identification result is invalid, the system uses the previous set of valid dynamic reference parameters, and maintains the real-time feed rate at the reference rate command value set by the operator until the identification trigger condition is met again.

[0025] In one specific embodiment of the present invention, before performing dynamic parameter identification, the system determines the current operating stage of the machine tool feed axis and isolates the spatial movement process of the tool without contacting the material from the cutting process. The parameter identification module establishes a time window triggering mechanism in conjunction with the feedforward data of the trajectory interpolator, specifically including the following sub-steps: S101, Parse the pre-read program code to determine the idle state. The trajectory interpolator performs a pre-read operation on the input CNC machining program, extracting the preparatory function instruction code of the current program segment to be executed. The system determines the type of the preparatory function instruction code. When the preparatory function instruction code is a rapid positioning instruction, it is determined that the feed axis is in an idle state.

[0026] When the preparation function instruction code is an interpolation run instruction, the system extracts the current coordinate variable of the tool's position along the spindle axis. This coordinate variable is compared with the tool safety height plane defined in the CNC machining program. If the current coordinate variable of the tool's position along the spindle axis is higher than the tool safety height plane, the system determines that the feed axis is in an idle state. If the current coordinate variable is not higher than the tool safety height plane, the system determines that the feed axis is in the cutting stage, blocks the trigger signal, and does not execute the parameter identification process.

[0027] The above comparison logic separates the non-cutting approximation process of the tool from the cutting feed process, providing a data cleaning basis for dynamic calibration.

[0028] S102 monitors the feed rate to filter out nonlinear dead zones. After confirming that the feed axis is in an idle state, the parameter identification module further extracts the commanded operating speed of the feed axis. The machine tool servo motor exhibits low-speed crawling during the start-stop phase near zero speed, and the process of converting static friction into dynamic friction on the guide rail has nonlinear characteristics. Directly collecting the power data during start-stop transients will lead to fitting errors in the subsequently established linear dynamic benchmark model.

[0029] The system presets effective speed thresholds in the parameter identification module. The effective speed thresholds include the effective speed threshold for the X-axis and the effective speed threshold for the Y-axis. Their values ​​are set as the critical stable operating speed required for the corresponding feed axis to overcome mechanical static friction. They are obtained based on the low-speed crawling test limit value of the motor calibrated by the servo drive factory, or determined based on the minimum operating speed when the speed fluctuation rate is lower than the preset stable threshold during the machine tool idle test.

[0030] The parameter identification module compares the absolute value of the command execution speed with the effective speed threshold in real time. When the absolute value of the command execution speed exceeds the effective speed threshold, the system determines that the feed axis has left the zero-speed nonlinear dead zone, and the parameter identification module generates a time window enable signal.

[0031] S103, based on the enable signal, opens a time window and establishes a discrete sampling sequence. After receiving the time window enable signal, the parameter identification module opens the window for a duration of... Dynamically identified time window. Duration The setting is based on the running time of the feed axis from standstill to the target speed. The setting range is 1.2 to 2 times the acceleration time of the system command, so as to ensure that the identification window covers the acceleration segment and the constant speed segment.

[0032] Within this dynamic identification time window, the system uses the synchronization period of the real-time communication bus as the time step and collects data at equal intervals. A total of discrete data points are used as sampling samples. To ensure that the overdetermined system of equations constructed using multiple sets of data has a unique least-squares numerical solution, the total number of sampling points is [number missing]. The numerical configuration is greater than the number of unknowns in the dynamic model to be identified. Furthermore, if the forward axis command speed drops to zero at the end of the dynamic identification time window, or the tool axial coordinate drops below the tool safety height plane, the parameter identification module determines that the acquisition process has been interrupted, the system clears the discrete sampling sequence of the current batch, and restarts the identification process when the trigger condition is met again.

[0033] After the dynamic identification time window is opened, the system starts sampling the instruction stream inside the CNC main control unit and the machine tool physical status data in parallel. The parameter identification module realizes the latching and retrieval of data in the same time dimension through the underlying bus protocol, which specifically includes the following sub-steps: S104, extract feedforward kinematic command parameters. During the preprocessing stage, the trajectory interpolator discretizes the continuous geometric trajectory into position increments for each interpolation cycle. The parameter identification module determines the total number of sampling points based on the time window. At each time step, feedforward kinematic data is read from the shared memory area of ​​the CNC main control unit. Let the current step be the first step within the identification time window. There are 1 sampling points, among which The value ranges from 1 to Positive integers.

[0034] The parameter identification module extracts the feedforward command speed of the machine tool's X-axis at this moment. Acceleration with feedforward instructions And extract the feedforward command running speed of the machine tool's Y-axis. Acceleration with feedforward instructions In practical implementation, if the underlying interface of the CNC main control unit only supports reading the feedforward command execution speed, the system uses the difference in command execution speed between two adjacent time step sizes divided by the time step size to obtain the feedforward command execution acceleration at the corresponding moment. The aforementioned feedforward kinematic command data is generated by the system interpolation algorithm and serves as the theoretical reference input variable for subsequently constructing the dynamic model.

[0035] The S105 performs servo electrical data synchronization sampling based on a distributed clock. The real-time communication bus is configured with a distributed clock synchronization mechanism at its underlying layer. The CNC master control unit, acting as the master node, broadcasts synchronization messages with global timestamp information to the bus network at fixed communication cycles. The X-axis and Y-axis servo drives, acting as slave nodes, generate internal interrupts based on the timestamps contained in the synchronization messages upon receiving them, triggering data sampling operations for the current and speed closed loops.

[0036] At each internal interrupt trigger moment, the servo driver measures the instantaneous torque current component and rotor mechanical angular velocity of the drive motor. The microprocessor inside the servo driver calls the preset torque constant of the servo motor and multiplies the instantaneous torque current component, torque constant, and rotor mechanical angular velocity to obtain the current no-load shaft-end mechanical output power of the motor. The no-load shaft-end mechanical output power is the shaft-end mechanical power calculated based on the servo current and rotor angular velocity, and does not include the driver's own electrical losses.

[0037] The parameter identification module reads the first parameter within the time window through the process data object mapping function of the real-time communication bus. The unloaded shaft end mechanical output power of the X-axis servo driver at each sampling point time. and the unloaded axis mechanical output power of the Y-axis servo driver. This hardware clock alignment mechanism ensures that the physical power data acquired by the servo's underlying layer is aligned with the feedforward motion commands generated by the interpolator on the time base.

[0038] S106, Construct a discrete sampling data set. The parameter identification module will be based on the same sampling point. The acquired feedforward kinematic command parameters and servo idle axis mechanical output power parameters are aggregated into discrete samples. For each independent axis system in the feed plane, the parameter identification module constructs the first... A sample vector.

[0039] For the X-axis, the first Each sample vector contains the feedforward instruction execution speed. Feedforward instruction execution acceleration and unloaded shaft end mechanical output power For the Y-axis, the first Each sample vector contains the feedforward instruction execution speed. Feedforward instruction execution acceleration and unloaded shaft end mechanical output power As time window progresses... The parameter identification module generates a total of dimensionality values ​​for each sampling point in the system memory through sequential execution. The X-axis and Y-axis empty-cut sample matrices are used to complete the acquisition and storage of empty-run travel data.

[0040] After acquiring a preset number of discrete sampling data samples, the parameter identification module performs matrix construction and algebraic solving to calculate the mechanical friction and inertial characteristics under the current machine tool operating environment. This includes the following sub-steps: S107, Construct a single-axis dynamic benchmark model. The parameter identification module establishes power balance equations for the independent axis system in the feed plane. In the non-cutting idle state, the mechanical output power of the servo motor's idle shaft end is used to overcome the equivalent mechanical resistance and inertial load of the transmission components. Taking the feed axis... For example, among which This is an identifier representing the X-axis or Y-axis. Within the time window... Mechanical output power of unloaded shaft end at each sampling point The theoretical equations satisfy: ; In the formula, For feed axis The dynamic equivalent friction coefficient; For feed axis The dynamic equivalent moment of inertia; This refers to the speed at which feedforward instructions are executed. Acceleration for feedforward instruction execution.

[0041] Based on the principles of mechanical system dynamics, this power balance equation consists of the superposition of steady-state dissipation and transient work done. Among them, The term characterizes the steady-state power consumed by the transmission components to overcome viscous frictional resistance; The term represents the transient dynamic power consumed to overcome the system's inertial load, which is the product of the inertial force and the operating speed.

[0042] S108, construct the parameter identification overdetermined matrix. Changes in the physical state during CNC machine tool operation can cause deviations in the fixed preset mechanical parameters. The parameter identification module retrieves the accumulated... A discrete data sample, for the feed axis Build dimensional observation matrix and dimensional power column vector Observation matrix The row elements are determined by the speed of feedforward instructions. Product of velocity and acceleration Composition; Power column vector The row element is the corresponding unloaded shaft end mechanical output power. Define the parameter vector to be identified as follows: It includes the dynamic equivalent friction coefficient. With dynamic equivalent moment of inertia The linear matrix equation established by the system is expressed as: ; S109, perform least squares regression and update the parameters. (Due to the total number of sampling points...) The configuration is set to be greater than the number of unknown parameters, and the observation matrix satisfies the full column rank condition. The aforementioned linear matrix equations belong to an overdetermined system of equations solvable using least squares regression. The parameter identification module uses the least squares regression algorithm to find the optimal parameter set that minimizes the sum of squared residuals. To prevent the lack of acceleration / deceleration characteristics in the sampled data from causing a reduction in matrix rank, the parameter identification module performs a pre-calculation analysis on the observation matrix. Perform a full-rank check.

[0043] If multicollinearity is detected in the data sequence, the system determines that the matrix is ​​singular, triggers the discard mechanism, and reuses the parameter values ​​from the previous identification to avoid computational overflow. Under the condition of full rank, the parameter identification module performs a pseudo-inverse operation on the above linear matrix equation using the underlying mathematical library to calculate the numerical solutions of the current parameter vectors to be identified on the X and Y axes.

[0044] The system calculates and obtains the dynamic equivalent friction coefficients and dynamic equivalent moments of inertia for each axis, then performs amplitude limiting and smoothing processing before writing them into the system register, completing the hot-state update of the operating parameters. The smoothing process uses a weighted update method, fusing the current identification results with the previous set of valid dynamic reference parameters according to preset weights to avoid abrupt changes in the dynamic reference parameters caused by a single sampling disturbance. The updated dynamic reference parameters serve as the reference data for separating the true cutting power components in the subsequent cutting stage.

[0045] After the machine tool feed axis enters the formal cutting stage, the power stripping module acquires the command flow information of the CNC system, providing theoretical reference data for stripping the pure cutting power component and monitoring mechanical deflection. This includes the following sub-steps: S201, execute cutting state determination and module activation. The power stripping module obtains the workpiece upper surface coordinates set by the machining program by reading the machine tool coordinate system offset parameters. The module monitors the current coordinates of the tool spindle axis through the position feedback data of the Z-axis servo drive or the Z-axis machine tool coordinate register of the CNC main control unit, and extracts the preparation function instruction code for the current interpolation cycle.

[0046] When the preparation function instruction code is a linear interpolation or circular interpolation instruction, and the current coordinate of the tool spindle axis is not higher than the coordinate of the workpiece's upper surface, the system determines that the feed axis has ended its idle process and the tool has entered the formal cutting phase. At this time, the system activates the calculation process of the power stripping module, synchronizing its running cycle with the interpolation cycle of the trajectory interpolator.

[0047] S202, calculates the tangential vector angle of the theoretical trajectory. In CNC machining, the direction of the command velocity vector in the current interpolation cycle is the tangential direction representing the geometric trajectory. The power stripping module calculates the geometric direction of the theoretical machining trajectory based on the command velocity components. Let the current step be the [number]th [section / phase] of the formal cutting segment. Each interpolation cycle, the system reads the feedforward command execution speed of the X-axis. Y-axis feedforward command execution speed To prevent division by zero errors in the calculation, the power stripping module calculates the composite velocity amplitude of the feed plane.

[0048] When the synthesis speed amplitude is greater than zero, the power stripping module calls the four-quadrant arctangent function to... and As input parameters, the theoretical tangential vector angle of the tool trajectory in the feed plane is calculated. This angle parameter characterizes the theoretical travel direction under conditions of no cutting resistance interference, serving as a benchmark for subsequent decoupling of the actual physical deflection state. When the synthesized velocity amplitude is zero, the system skips the function calculation and assigns the theoretical trajectory tangential vector angle of the current interpolation cycle to the calculation result of the previous interpolation cycle.

[0049] S203, extracts real-time feedforward kinematic data. This provides input conditions for calculating the non-cutting power dissipation of the current cycle. The power stripping module is in the first... Within each interpolation cycle, the module synchronously extracts the kinematic control commands for the feed plane. It also extracts the X-axis feedforward commands and acceleration from the data sharing area of ​​the trajectory interpolator. Y-axis feedforward command running acceleration The acquired velocity and acceleration parameters are aligned temporally with the tangential vector angle of the theoretical trajectory, forming a benchmark dataset characterizing the feed state of the current interpolation cycle. This dataset is stored in the system cache for subsequent input into the dynamic benchmark model to perform loss calculations.

[0050] In one specific embodiment of the present invention, during the cutting feed along the workpiece contour, the total output power of the servo motor is used not only to remove the cutting material, but also to drive the moving parts of the machine tool itself to overcome friction and inertia. To obtain purely physical characteristics reflecting the cutting state, the power stripping module calculates the basic mechanical dissipation under the current feed state in real time, specifically including the following sub-steps: S204, retrieve dynamic reference parameters. The power stripping module is in... Within each interpolation cycle, the system registers of the CNC main control unit are accessed. The module extracts the dynamic equivalent friction coefficient of the machine tool X-axis identified during the idle run phase. With dynamic equivalent moment of inertia and the dynamic equivalent friction coefficient of the machine tool's Y-axis With dynamic equivalent moment of inertia These parameters characterize the basic mechanical properties of the machine tool under its current operating environment.

[0051] S205 calculates the power dissipation of a single-axis non-cutting machine. The power stripping module substitutes the extracted dynamic reference parameters and feedforward kinematic data into the dynamic equations. The system independently calculates the non-cutting power dissipation of each axis in the feed plane. For the X-axis, its first... Non-cutting power dissipation per interpolation cycle The calculation equation is as follows: ; For the Y-axis, its first Non-cutting power dissipation per interpolation cycle The calculation equation is as follows: ; In the formula, and The running speed of the marked feedforward command for the corresponding shaft system; and This is the absolute value of the execution speed of the corresponding feedforward instruction; and The feedforward command acceleration for the corresponding shaft system.

[0052] The work done by the moving parts of the machine tool against friction is always manifested as the energy consumption of the servo system. Therefore, the friction dissipation term is calculated using the absolute value of velocity. The inertial work term retains the product of acceleration and signed velocity to reflect the difference in power direction when the feed axis accelerates and decelerates. The absolute value of velocity is introduced to ensure that the friction dissipation term is always positive power. The sign of the inertial work term is determined by the relative direction of acceleration and velocity. Using command feedforward data instead of sensor feedback data for calculation eliminates mechanical transmission disturbances caused by cutting forces during machining, allowing the calculated power dissipation to characterize the inherent losses of the machine tool running along its theoretical trajectory.

[0053] S206, performs power dissipation sign determination and timing alignment. In CNC trajectories with alternating acceleration and deceleration, the product of dynamic equivalent inertia and acceleration produces alternating positive and negative power values. When the feed axis is accelerating, the servo motor outputs power to overcome inertia and do positive work; when the feed axis is decelerating, the moving parts release kinetic energy, and the inertial power term is negative. The power stripping module retains the algebraic sign of the calculated power dissipation and... , It is bound to the system timestamp of the current interpolation period. The timestamped power dissipation data is written to the data buffer as a reference subtrahend for subsequent stripping operations.

[0054] See attached document Figure 2 After calculating the basic mechanical dissipation in real time, the system performs actual electrical sampling and differential calculation to isolate the power data reflecting the interaction between the tool and the workpiece. This includes the following sub-steps: S207, synchronously acquire the mechanical output power of the shaft end in the cutting state. In the first stage of the formal cutting phase... Within each interpolation cycle, the real-time communication bus triggers an internal interrupt of the servo driver based on a global timestamp. The servo driver calculates the mechanical output power of the motor's cutting-state shaft end during that interpolation cycle based on the instantaneous torque current component, the servo motor torque constant, and the rotor mechanical angular velocity. The power stripping module reads the X-axis cutting-state shaft end mechanical output power via the bus. and the mechanical output power of the Y-axis cutting state shaft end The physical measurement process is synchronized with the feedforward kinematic command extraction process on a time base.

[0055] S208 performs power differential decoupling. The power stripping module subtracts the feedforward theoretical dissipation calculation results from the actual electrical sampling results for the same timestamp. By subtracting the corresponding non-cutting dissipation power from the cutting-state mechanical output power of the X-axis and Y-axis respectively, the pure cutting power components of the X-axis and Y-axis in the current interpolation cycle are obtained. and .

[0056] This calculation process removes frictional losses and inertial work from the total energy consumption of the machine tool transmission components. The remaining power component represents the mechanical work done to overcome material shear deformation. During the calculation, if the calculated pure cutting power component is negative due to model identification residuals, sudden changes in coolant fluid resistance, or servo sampling noise, the power stripping module sets a lower limit threshold of zero for the calculation result to limit its amplitude. This is to comply with the physical constraint that the work done in material removal is always positive and to prevent abnormally negative values ​​from being passed on to subsequent algorithms.

[0057] Meanwhile, the system records the number of consecutive cycles in which negative values ​​occur and the magnitude of the negative values; when the duration of a negative value exceeds the preset number of cycles, it determines that there is a deviation in the current dynamic benchmark model and triggers re-identification or reverts to benchmark operation.

[0058] S209, Discrete low-pass filtering of pure cutting power component data. Due to the operation of the current loop switch inside the servo system and the periodic excitation caused by the intermittent cutting of the material by the multi-bladed tool, the pure cutting power component signal obtained by differential calculation is mixed with high-frequency interference. To avoid broadband noise causing abrupt changes in the subsequent deflection angle calculation results, the power stripping module... and Perform a first-order discrete low-pass filter. Taking the X-axis as an example, its digital filtering equation is: ; In the formula, The filtered first The pure cutting power component of the X-axis in one interpolation cycle; This is the filtered pure cutting power component of the X-axis from the previous cycle. This is the first-order filter coefficient, with a value ranging from 0 to 1. To achieve adaptive adjustment of the filtering characteristics, the power stripping module reads the spindle speed and the number of cutting edges set in the machining program, multiplies the spindle speed by the effective number of cutting edges, the number of grinding stages, or the equivalent periodic excitation parameter, and divides by 60 to calculate the periodic excitation reference frequency in Hz.

[0059] The power stripping module uses a preset proportion of the excitation reference frequency of this cycle as the cutoff frequency of the first-order low-pass filter. The preset proportion is between 0.2 and 0.5. Combined with the inherent interpolation period of the trajectory interpolator, the aforementioned first-order filter coefficients are calculated. By setting the cutoff frequency below the periodic excitation reference frequency, the filter can reduce high-frequency components caused by intermittent contact of the tool edge, periodic impact of abrasive grains, or disturbance of grinding debris, while retaining the power envelope that reflects the slow changes in cutting load.

[0060] The aforementioned dynamic tuning mechanism ensures that the filter can remove the excitation components caused by multi-edge intermittent cutting under different cutting conditions, while retaining the envelope of cutting torque changes. The pure cutting power component of the Y-axis is filtered using the same structure and coefficients. The system then filters the pure cutting power component of the X-axis. Compared with the pure cutting power component of the Y-axis Output to a buffer for subsequent calculation of the actual resistance direction.

[0061] After extracting and smoothing the pure cutting power component data, the deflection decoupling module calculates the spatial orientation of the actual cutting resistance in the feed plane based on the physical mapping relationship between work done and motion state. This includes the following sub-steps: S301, Feed Rate Status Assessment and Overflow Protection. The solution for cutting resistance depends on the quotient of power and speed. To prevent division-by-zero overflow anomalies caused by the motion speed approaching zero at dwell points or sharp corner reversal points in the CNC machining trajectory, the deflection decoupling module adjusts the X-axis feedforward command speed of the current interpolation cycle. Y-axis feedforward command execution speed The absolute value is checked.

[0062] The module calls a micro-speed determination threshold, which is calculated based on the position feedback resolution of the machine tool servo system and the interpolation cycle of the trajectory interpolator. Its setting range is 0.001 mm / s to 0.01 mm / s. When the absolute value of the speed of a certain feed axis is less than the micro-speed determination threshold, the system determines that the axis is in a zero-speed standstill or low-speed reversal state. Instead of directly setting the current cutting resistance component of the axis to zero, the system reads the torque current or equivalent disturbance torque data of the axis servo drive, subtracts the friction torque and inertial torque calculated by the dynamic reference model, obtains the cutting disturbance torque of the axis, and converts it into the cutting resistance component of the axis based on the transmission ratio, lead screw lead, or equivalent force conversion coefficient.

[0063] If the current servo drive does not support disturbance torque reading, the cutting resistance component of this axis will remain the calculation result of the previous effective interpolation cycle. When the absolute value of the speed is greater than or equal to the micro-speed determination threshold, the corresponding resistance component mapping calculation will be triggered.

[0064] S302, Single-axis cutting resistance component mapping calculation. The deflection decoupling module calculates the cutting resistance component using the principle of mechanical work. For feed axes that meet the micro-speed judgment threshold, when the absolute value of the feed axis speed is greater than or equal to the micro-speed judgment threshold, the deflection decoupling module divides the filtered pure cutting power component by the corresponding feedforward command running speed and introduces a direction correction mechanism to obtain the cutting resistance component.

[0065] When the absolute value of the feed axis speed is less than the micro-speed judgment threshold, the deflection decoupling module uses the servo disturbance torque of that axis to calculate the cutting resistance component. This is to avoid power division overflow due to the speed approaching zero, and to prevent the stationary bearing from being misjudged as having no resistance when subjected to lateral cutting forces. X-axis cutting resistance component. Y-axis cutting resistance component The mapping calculation equation is: ; ; In the formula, the negative sign represents the actual cutting resistance as a reaction force of the workpiece material on the tool, and its vector direction is opposite to the direction of the tool feed. The above calculation transforms the pure cutting power component into a mechanical component with algebraic signs by combining the commanded operating speed with directional characteristics.

[0066] S303, actual cutting resistance vector angle synthesis. After obtaining the cutting resistance components of the orthogonal biaxial feed plane, the deflection decoupling module reconstructs the geometric orientation of the cutting resistance. The deflection decoupling module calls the four-quadrant arctangent function to determine the Y-axis cutting resistance component. and X-axis cutting resistance component The actual cutting resistance vector angle for the current interpolation cycle is calculated using the input parameter. This angular parameter characterizes the direction of two-dimensional resistance acting on the tool tip after being affected by factors such as tool wear, material anisotropy, and chip extrusion.

[0067] If the cutting resistance components of the X and Y axes are both invalid during the current interpolation cycle, it indicates that the tool has no effective displacement in the feed plane. The deflection decoupling module skips the arctangent calculation and uses the current actual cutting resistance vector angle. The calculated value is maintained at the level of the previous effective interpolation cycle to ensure the temporal continuity of the angle data. If the system enters the cutting segment for the first time and there is no actual cutting resistance vector angle from the previous effective interpolation cycle, then... Initialize it to the theoretical resistance reference angle to prevent logical breaks in subsequent angle deviation calculations.

[0068] After reconstructing the actual cutting resistance vector angle, the system compares it with the theoretical trajectory geometric reference to extract feature data characterizing the tool's force deflection state and its evolution trend. This includes the following sub-steps: S304, Phase Deviation Calculation and Periodic Normalization. Under ideal machining conditions, the direction of the cutting resistance is collinear with and opposite to the direction of the tool feed rate. To obtain the angular reference, the deflection decoupling module extracts the X-axis feedforward command speed and Y-axis feedforward command speed of the current interpolation cycle, and calculates the theoretical trajectory tangential vector angle using the arctangent function. The deflection decoupling module adds the constant pi to the tangential vector angle of the theoretical trajectory. As the theoretical resistance reference angle, the actual cutting resistance vector angle of the current interpolation cycle is calculated. The difference between the original phase difference and the theoretical resistance reference angle is used to obtain the original phase difference value.

[0069] To address the periodicity of angle data and eliminate numerical jumps when crossing coordinate axis boundaries, the deflection decoupling module performs boundary checks and translation processing on the original phase difference: when the original phase difference is greater than... When, subtract When the original phase difference is less than At that time, plus ; in and When the time interval is between, it remains unchanged. The processed output is a normalized phase deviation, the numerical range of which is limited to... Within the range.

[0070] Before determining phase deviation and micro-deflection polarity factor, the system establishes a unified geometric reference within the feed plane. A right-handed plane coordinate system is constructed using the positive X-axis and positive Y-axis directions of the machine tool, with the theoretical trajectory tangential vector angle as the reference. The corresponding direction is used as the theoretical feed tangential direction of the tool, with This serves as the theoretical resistance reference direction. A normalized phase deviation greater than zero indicates that the actual cutting resistance vector deflects counterclockwise relative to the theoretical resistance reference direction, while a normalized phase deviation less than zero indicates that the actual cutting resistance vector deflects clockwise relative to the theoretical resistance reference direction.

[0071] S305, Absolute Phase Deviation Extraction. The tool deflection state depends on the magnitude of the spatial angle deviating from the ideal cutting state, and is independent of the algebraic sign of the left or right side of the deflection trajectory. The deflection decoupling module performs an absolute value operation on the normalized phase deviation to obtain the absolute phase deviation of the current interpolation cycle. The unit of this parameter is rad. It eliminates the interference of coordinate system directionality and is used to characterize the magnitude of the deviation of the cutting-edge force from the theoretical state.

[0072] S306, Sliding window rate of change extraction. To capture the dynamic evolution characteristics of the tool deflection state over time, the deflection decoupling module constructs a length of [missing information] in system memory. A first-in, first-out (FIFO) data buffer window. The deflection decoupling module will determine the absolute phase deviation of the current interpolation cycle. The data is pushed into the front of the buffer window, and the oldest historical period data within the window is simultaneously removed. Using the discrete data sequences at the beginning and end stored in the buffer window, the deflection decoupling module performs differential calculations to extract the sliding window rate of change of the absolute phase deviation. The calculation equation is as follows: ; In the formula, The absolute phase deviation of the oldest valid interpolation period stored within the data cache window; The length of the sliding data window is configured with parameters that are positive integers greater than or equal to 2, typically set to cover a physical time span of 0.05s to 0.2s. Those skilled in the art can calculate this based on the specific interpolation cycle of the trajectory interpolator. The specific value; The interpolation period of the trajectory interpolator is expressed in seconds. The calculated sliding window change rate is expressed in rad / s. This rate of change reflects the direction and speed of change of the absolute phase deviation over time, providing a parameter basis for subsequent construction of control intervention logic. (Time-stamped data is included.) and It is packaged and transferred to the system's shared storage area.

[0073] See attached document Figure 3 After extracting the absolute phase deviation and rate of change, the deflection decoupling module combines the spatial motion properties of the tool to determine the positional relationship between the micro-force deflection of the tool tip and the machining contour, thereby identifying the machining error trend caused by the deflection. Specifically, it includes the following sub-steps: S307, Machining contour topology extraction. The deflection of cutting forces causes overcutting or undercutting on the workpiece surface, depending on the spatial position of the tool relative to the workpiece cutting edge. The deflection decoupling module reads the tool radius compensation status command for the current interpolation cycle. This command includes left tool compensation commands, right tool compensation commands, and a no-compensation state command to cancel tool radius compensation.

[0074] The tool left compensation command indicates that the tool center trajectory is located to the left of the workpiece contour along the tool feed direction; the tool right compensation command indicates that the tool center trajectory is located to the right of the workpiece contour; the no-compensation state indicates that the tool center moves along the programmed trajectory without specifying an offset side. The deflection decoupling module uses the extracted compensation state as the spatial geometric reference for subsequent polarity mapping.

[0075] S308, Determination of resistance deflection direction. The deflection decoupling module calls the normalized phase deviation data and extracts its algebraic symbol. This algebraic symbol represents the two-dimensional rotation direction of the actual cutting resistance vector relative to the theoretical resistance reference angle.

[0076] When the normalized phase deviation is greater than zero, the deflection decoupling module determines that the actual cutting resistance vector has deflected counterclockwise relative to the reference; when the normalized phase deviation is less than zero, it determines that the actual cutting resistance vector has deflected clockwise. This step converts continuous angle data into discrete direction state parameters.

[0077] S309, Microscopic deflection polarity factor mapping. The deflection decoupling module combines the machining contour topology and the resistance deflection direction for cross-determination, and outputs the microscopic deflection polarity factor.

[0078] The deflection decoupling module determines the workpiece solid side normal direction based on the tool radius compensation state. In the left-compensated tool state, observing along the theoretical feed direction, the workpiece solid side normal direction is located to the right of the tool center trajectory. With the tool in right-compensated state, when viewed along the theoretical feed direction, the normal direction on the workpiece solid side is located to the left of the tool center trajectory. The deflection decoupling module projects the deflection component of the actual cutting resistance vector relative to the theoretical resistance reference direction onto the normal direction on the workpiece solid side. If the projection result points into the workpiece solid, it is determined that the tool has an overcutting tendency to squeeze into the workpiece solid, and the micro-deflection polarity factor is assigned a value of 1; If the projection result deviates from the interior of the workpiece, it is determined that the tool has an undercutting tendency away from the workpiece, and the micro-deflection polarity factor is assigned a value of -1. If the normalized phase deviation of the current interpolation cycle is equal to zero, the projection result is lower than the preset polarity recognition threshold, or the system is in an uncompensated state, the micro-deflection polarity factor is assigned a value of 0.

[0079] During the aforementioned determination process, if the normalized phase deviation of the current interpolation cycle is zero, or the system is in an uncompensated state, the deflection decoupling module assigns a value of 0 to the micro-deflection polarity factor, indicating that no lateral deflection machining error with a definite cutting edge has been identified. The extracted micro-deflection polarity factor is a dimensionless constant that transforms the geometric angle deflection into a physical machining state identifier, used to determine the direction of feed rate compensation increase or decrease in subsequent control algorithms. The timestamped micro-deflection polarity factor data is written to the shared storage area for subsequent modules to access.

[0080] After acquiring real-time data characterizing the force deflection of the cutting tool, the system establishes an evaluation benchmark to quantify the boundary of the machining state, and performs a conventional feed maintenance action when the machine tool is determined to be in a stable cutting phase. This includes the following sub-steps: S401, Determination of physical thresholds for control intervention. To distinguish between normal cutting force fluctuations and abnormal deflections that cause machining deviations, the compensation control module presets an absolute phase deviation tolerance threshold and a sliding window change rate tolerance threshold.

[0081] The threshold is set based on the tool diameter, tool overhang length, glass material hardness, fracture toughness, elastic modulus, allowable chipping width, edge roughness requirements, and critical material removal parameters of the current machining task, and is calculated by combining the calibration data of the pre-cutting experiment or by querying the machine tool's preset machining process database.

[0082] In the CNC contour machining of glass cover plates, the absolute phase deviation tolerance threshold is set between 0.05 rad and 0.15 rad, and the sliding window change rate tolerance threshold is set between 0.1 rad / s and 0.4 rad / s. The determined threshold parameters are written into the CNC system's storage area as a comparison reference for switching cutting states.

[0083] S402, Cutting State Cross-Determination. The compensation control module reads the absolute phase deviation and sliding window change rate in the shared storage area in real time during each interpolation cycle, and compares them numerically with preset tolerance thresholds. This comparison operation comprehensively evaluates the transient amplitude and dynamic change rate of the tool force deflection to determine whether the current machining state exceeds the process allowable limits.

[0084] S403, reference maintenance strategy under normal cutting conditions. When the compensation control module determines that the absolute phase deviation of the current interpolation cycle is less than or equal to the absolute phase deviation tolerance threshold, and the sliding window change rate of the current interpolation cycle is less than or equal to the sliding window change rate tolerance threshold, the system determines that it is currently in a normal cutting state.

[0085] Meeting the above criteria indicates that the deformation and trend of the tool under cutting resistance are within the allowable range of the process and do not pose a risk of causing defects in the machined surface quality. In this state, the compensation control module does not trigger the dynamic intervention mechanism for the feed rate, generates a normal maintenance command, and maintains the spindle speed ratio at the reference spindle speed ratio set by the operator. The compensation control module accesses the CNC system's human-machine interface register to read the reference ratio command value set by the operator on the CNC machine tool's operating panel, and maintains the real-time feed rate of the current interpolation cycle at this reference ratio command value. The compensation control module sends this reference ratio command value to the trajectory interpolator in the CNC core. The trajectory interpolator, based on this command value and the nominal feed rate set by the machining code, plans the position increment of each feed axis in subsequent interpolation cycles. This strategy ensures that the machine tool processes at the set efficiency under normal operating conditions.

[0086] See attached document Figure 4 During the cross-cutting state determination process, if the machining state is detected to exceed the allowable process boundary and the tool is pressed into the workpiece, the system triggers a dynamic deceleration compensation mechanism, which includes the following sub-steps: S404, Overcutting State Trigger Detection. The compensation control module jointly compares the state parameters of the current interpolation cycle with the micro-deflection polarity factor. If the absolute phase deviation of the current interpolation cycle is greater than the absolute phase deviation tolerance threshold, or if the sliding window change rate of the current interpolation cycle is greater than the sliding window change rate tolerance threshold, it indicates that the tool tip force deflection deformation deviates from the allowable range of the process.

[0087] Based on this, the compensation control module reads the micro-deflection polarity factor from the shared storage area. If the micro-deflection polarity factor is assigned a value of 1, the system determines that the tool is deflected into the workpiece under the lateral reaction force, and is in an overcutting state of inward micro-extrusion. This state is prone to causing local dimensional deviations in the workpiece, triggering the intervention logic.

[0088] S405, Feed rate attenuation coefficient calculation. For overcutting conditions, the compensation control module reduces the material removal load per unit time by decreasing the tool feed rate, thereby reducing the glass material removal rate, grinding contact load, and the risk of edge chipping. It passively overcomes the force deflection by relying on the structural rigidity of the machine tool spindle and tool holder. The system calculates the feed rate attenuation coefficient using the excess of the absolute phase deviation and the sliding window change rate. The specific calculation process is as follows: The system calculates the first difference between the absolute phase deviation of the current interpolation period and the absolute phase deviation tolerance threshold, and the second difference between the sliding window change rate of the current interpolation period and the sliding window change rate tolerance threshold. When the first difference is less than zero, the first difference is set to zero; when the second difference is less than zero, the second difference is set to zero.

[0089] Subsequently, the system multiplies the first difference by a preset deviation penalty weight to obtain a first product, and multiplies the second difference by a preset rate of change penalty weight to obtain a second product. Finally, the first product and the second product are added together to obtain the feed rate decay coefficient for the current interpolation cycle.

[0090] In the above calculation, the deviation penalty weight quantifies the impact of amplitude deviation on the multiplier, and its value range is set between 0.5 and 1.5; The rate-of-change penalty weight quantifies the impact of the rate of change on the multiplier, and its value is set between 0.2 and 0.8. The specific value of the weight parameter is based on the response bandwidth calibration of the machine tool servo drive. When the servo system has a wide response bandwidth, a larger weight is applied to improve dynamic compensation sensitivity; conversely, a smaller weight is applied to avoid control oscillations in the feed axis. The calculated feed rate attenuation coefficient is a dimensionless proportional parameter of the same order of magnitude as the reference rate command value.

[0091] S406, Coordinated speed reduction execution and lower limit clamping. The compensation control module subtracts the calculated feed rate attenuation coefficient from the operator-set baseline feed rate command value to obtain the intermediate feed rate. To prevent the machine tool from triggering interpolation axis stoppage due to excessive speed reduction, or from causing micro-crack propagation, localized thermal damage, or deterioration of edge roughness on the workpiece surface due to cutting edge slippage relative to the material, the compensation control module performs lower limit clamping processing on the intermediate feed rate.

[0092] The system presets a minimum safety margin threshold, which is set to 30% to 50% of the base margin command value. When the intermediate margin is lower than the minimum safety margin threshold, the system assigns the minimum safety margin threshold as the real-time feed margin; when the intermediate margin is greater than or equal to the minimum safety margin threshold, the system uses the intermediate margin as the real-time feed margin.

[0093] The compensation control module sends the real-time feed rate after the lower limit clamping to the trajectory interpolator. The trajectory interpolator uses this real-time feed rate to synchronously adjust the interpolation step size of the X and Y axes, achieving coordinated speed reduction along the orthogonal axes at the same ratio. This coordinated speed reduction mechanism suppresses the continuous surge of cutting force and mitigates the risk of overcutting the workpiece without changing the nominal machining trajectory space.

[0094] See attached document Figure 5 During the cutting state determination process, if the machining state is detected to exceed the process allowable boundary and the tool retracts away from the workpiece, the system determines that the chip removal is currently obstructed and triggers the single-axis slow-release control mechanism, which specifically includes the following sub-steps: S407, Undercutting Condition Trigger Detection. The compensation control module verifies and compares the feature data of the current interpolation cycle. If the absolute phase deviation of the current interpolation cycle is greater than the absolute phase deviation tolerance threshold, or the sliding window change rate is greater than the sliding window change rate tolerance threshold, the system reads the micro-deflection polarity factor in the shared storage area. If the micro-deflection polarity factor is assigned a value of -1, the system determines that the tool is affected by glass abrasive blockage, poor coolant drainage, tool abrasive passivation, local hard inclusions, or transient edge fracture loads, resulting in outward undercutting deformation. This condition increases the risk of tool breakage and triggers the corresponding mitigation control logic.

[0095] S408, motion master axis identification. To release chip extrusion stress, the compensation control module adopts an asymmetric feed intervention mode. The compensation control module accesses the system interpolation preprocessing buffer and extracts the nominal X-axis feed increment and nominal Y-axis feed increment planned by the current interpolation cycle trajectory interpolator.

[0096] The system performs an absolute value operation on the nominal feed increments of the X-axis and Y-axis and compares their values. The feed axis with the larger absolute value is determined as the dominant motion axis of the current trajectory, and the feed axis with the smaller absolute value is determined as the passive holding axis. When the absolute values ​​of the nominal feed increments of the X-axis and Y-axis are equal, the system defaults to designating the X-axis as the dominant motion axis and the Y-axis as the passive holding axis. This step establishes the feed direction under the stress concentration of compression.

[0097] S409, Single-axis slow-release feed calculation and execution. The compensation control module applies a feed limit to the main motion axis individually, causing the tool to generate a transient avoidance trajectory. The specific calculation process is as follows: the system extracts a preset slow-release attenuation coefficient, multiplies the nominal feed increment of the main motion axis in the current cycle by this slow-release attenuation coefficient, and obtains the actual interpolation increment of the main motion axis after slow-release control. All increments mentioned above are in millimeters (mm). The slow-release attenuation coefficient is a dimensionless parameter, with a value range set between 0.1 and 0.3, and the specific value is determined based on the tool diameter and chip flute volume calibration.

[0098] After the calculation is completed, the compensation control module locks the actual interpolation increment of the driven holding axis to its original nominal feed increment, and updates the actual interpolation increment of the motion-dominant axis to the calculated actual interpolation increment. The module then sends the updated interpolation increment command for each axis to the corresponding servo drive position loop.

[0099] Because the feed displacement ratio of the orthogonal dual axes changes in the current interpolation cycle, the actual tool trajectory deviates locally from the nominal contour tangent, shifting towards the spatial direction corresponding to the driven holding axis. To prevent the mitigation action from introducing unacceptable contour errors, the compensation control module limits the trajectory offset of a single interpolation cycle and the cumulative offset of consecutive mitigation cycles, ensuring that they do not exceed a preset proportion of the allowable contour error value. This preset proportion ranges from 10% to 30%. When the cumulative offset reaches the limit value, the system stops further increasing the mitigation offset, maintaining only the current safety ratio or resuming synchronous interpolation.

[0100] This non-proportional displacement adjustment changes the transient force direction in the contact area between the cutting edge of the tool and the edge of the glass cover plate, reduces the continuous compression of grinding debris in the contact area between the tool and the workpiece, disrupts the grinding debris accumulation state, and promotes the discharge of grinding debris with the coolant.

[0101] In subsequent interpolation cycles, the compensation control module continuously monitors the absolute phase deviation and the sliding window change rate. When both parameters are less than or equal to the corresponding tolerance threshold, the system determines that the wear debris extrusion or local abnormal resistance has been released, stops applying the slow-release attenuation coefficient, and the trajectory interpolator resumes the synchronous interpolation action of the orthogonal dual axes.

[0102] During the recovery process, the compensation control module generates a reverse compensation increment based on the recorded cumulative trajectory offset, and smoothly superimposes it onto the motion-dominant axis or the driven-maintaining axis in multiple consecutive interpolation cycles, so that the tool center trajectory gradually returns to the nominal machining trajectory, avoiding secondary impact or contour vibration caused by single-cycle abrupt changes.

[0103] S410, Spindle speed ratio coordinated adjustment. While performing feed rate adjustment or single-axis slow-release control, the compensation control module reads the spindle load rate, actual spindle speed, and the trend of the filtered pure cutting power component from the spindle frequency converter driver. When the absolute phase deviation exceeds the absolute phase deviation tolerance threshold, and the pure cutting power component shows an upward trend over multiple consecutive interpolation cycles, the compensation control module determines that the load in the tool-glass cover contact area is increasing and generates spindle speed ratio adjustment parameters. These parameters are centered on the operator-set reference spindle speed ratio and vary between a preset upper limit and a preset lower limit. The preset upper limit is 110% to 120% of the reference spindle speed ratio, and the preset lower limit is 80% to 90% of the reference spindle speed ratio.

[0104] In overcutting conditions, the compensation control module prioritizes reducing the real-time feed rate and maintains the spindle speed ratio at the reference spindle speed ratio. In undercutting conditions or when chip removal is obstructed, if the spindle load rate does not exceed the preset spindle safety load threshold, the compensation control module adjusts the spindle speed ratio between the preset upper limit and the preset lower limit based on the trend of the filtered pure cutting power component, thereby enhancing the chip removal capability of the tool contact area.

[0105] If the spindle load rate exceeds the preset spindle safety load threshold, the compensation control module will prevent further increases in the spindle speed ratio and switch the control strategy to reduce the feed rate.

[0106] Through the aforementioned control process, the system can separate the machine tool's own friction and inertial load from the actual material removal load during the glass cover contour machining process, and identify the tool force deflection trend based on the offset of the cutting resistance direction relative to the theoretical trajectory direction. When the system detects that the tool is pressing against the workpiece solid side, it suppresses edge overcutting and chipping by reducing the feed rate; when the system detects that the tool is retracting away from the workpiece solid side, it releases the chip pressing state through single-axis slow release, trajectory recovery, and coordinated adjustment of spindle speed, thereby reducing the risk of undercutting, edge micro-crack propagation, and abnormal tool wear, and improving the dimensional stability and edge quality of the glass cover machining.

[0107] Specific application examples: The CNC adaptive feed machining control system and method for glass cover plates of the present invention establishes a dynamic reference in the non-cutting stroke, extracts the pure cutting power component by real-time stripping of non-cutting dissipated power during the cutting stage, and then reconstructs the actual cutting resistance vector angle and calculates the absolute phase deviation and microscopic deflection polarity factor. Based on the amplitude deviation and sliding window change rate of the above-mentioned deflection characteristics, the system judges the overcutting or undercutting trend of the tool in the microscopic space in real time, and adjusts the feed rate accordingly, performs single-axis slow-release intervention, or coordinates the spindle speed adjustment.

[0108] Experimental results and application feedback show that the system and control method can effectively suppress the sudden change in cutting force caused by material anisotropy or chip removal obstruction during glass contour processing without adding external mechanical sensors, control the contour error of the workpiece within a strict process tolerance zone, and reduce the chipping defects of brittle glass edges.

[0109] To illustrate the specific calculation and derivation process of this invention, we will use the machining of a 45-degree straight chamfer on the outer contour of a 0.7mm thick high-aluminosilicate glass cover plate as an example. The system is set to right-tool compensation mode, and the interpolation cycle of the trajectory interpolator is... Set to 0.001s. In the first... During each interpolation cycle, the system synchronously acquires the cutting-state mechanical output power of the X and Y axes and calculates non-cutting mechanical dissipation using dynamic reference parameters. After differential stripping and filtering, the pure cutting power components of each axis are obtained. To calculate the actual force deflection, the system uses the cutting resistance component mapping formula based on the mechanical work principle. In the formula This represents the cutting resistance component along the X-axis. The filtered first The pure cutting power component of the X-axis per interpolation cycle This refers to the speed at which the X-axis feedforward command runs.

[0110] The system performs arctangent calculations on the resistance components of each axis to reconstruct the actual cutting resistance vector angle, and calculates the absolute phase deviation by subtracting it from the theoretical resistance reference angle. When the machining operation reaches around the 250th interpolation cycle, the tool encounters local hard points and wear debris accumulation, causing an abnormal increase in the absolute phase deviation.

[0111] To assess the evolutionary trend, the system calls the sliding window rate of change extraction formula: In the formula The rate of change of the sliding window. This represents the absolute phase deviation of the current interpolation cycle. The absolute phase deviation of the oldest valid interpolation period stored within the data cache window. The window length is given. Calculations show that the sliding window change rate is greater than the preset tolerance threshold. The system, combined with the microscopic deflection polarity factor, determines that the chip removal is obstructed and the cutting is undercut. Then, according to closed-loop logic, the nominal feed increment of the motion dominant axis in the current interpolation cycle is multiplied by the slow-release attenuation coefficient to encourage the tool to smoothly avoid the chip and break the chip compression balance.

[0112] To verify the effectiveness of the control scheme, a comparative machining experiment was conducted on a CNC machine tool using the aforementioned high-alumina-silicate glass cover plate. (See attached diagram.) Figure 6 , Figure 6 The horizontal axis represents the machining time in seconds; the vertical axis represents the resultant force of the pure cutting power component in watts (W).

[0113] The dotted black lines represent the conventional machining mode without adaptive control, while the solid black lines represent the experimental group using the adaptive control method of this invention. As can be seen from the clear and smooth curve trends in the figure, the power curve of the conventional machining mode exhibits a smooth, half-sine-shaped surge peak with an amplitude approaching 180W when encountering material resistance in the 1.0s to 1.15s range. In contrast, the solid line representing the pure cutting power component of the experimental group, under the same operating conditions, only produces a small and gentle bulge through adaptive dynamic speed reduction and single-axis slow-release intervention. This bulge is then quickly and stably clamped within the nominal load safety limit of 115W, effectively filtering out high-frequency stray noise and suppressing transient stress deterioration.

[0114] See attached document Figure 7 This reflects the closed-loop tracking effect of the system on the microscopic deflection state of the tool throughout the entire stroke. Figure 7 The horizontal axis represents the sequence number of consecutive interpolation periods, and the vertical axis represents the calculated absolute phase deviation value in rad. The pure black dashed line represents the system's preset absolute phase deviation tolerance threshold of 0.05 rad, and the pure black solid line represents the real-time absolute phase deviation.

[0115] The measured noise reduction curves show that during normal cutting, the solid line fluctuates slightly around the 0.015 rad baseline. As it approaches the 250th interpolation cycle, due to a sudden increase in lateral resistance, the solid line smoothly and steeply crosses the dashed threshold, reaching a peak of 0.26 rad in the 250th cycle. Thanks to the underlying clock synchronization mechanism, the system generates a feed release command the instant the limit is exceeded, and the curve after intervention quickly and smoothly returns to the safe range within the following 3 to 4 cycles.

[0116] See attached document Figure 8 The dimensions and surface quality indicators of the final product were compared. Figure 8 The horizontal axis is divided into two independent evaluation categories: the root mean square value of the contour error and the maximum chipping width; the vertical axis represents the measured geometric dimensions in mm. To avoid visual obstruction, the upper limit of the Y-axis scale is specially raised to 0.035 mm. Each category in the figure contains two sets of pure grayscale bars: dark gray blocks represent the conventional processing mode, and light gray blocks represent the adaptive processing mode of this invention. Quantitative data shows that under the conventional mode, the root mean square value of the glass cover's contour error is 0.011 mm, and the maximum chipping width is 0.025 mm; after adopting this scheme, the root mean square value of the contour error in the experimental group is compressed to 0.003 mm, and the maximum chipping width is strictly controlled within 0.008 mm.

Claims

1. A CNC adaptive feed machining control method for glass cover plates, characterized in that, Includes the following steps: The mechanical output power and feedforward kinematic commands of the X-axis and Y-axis servo drives at the unloaded shaft ends are obtained during the non-cutting idle stroke. The dynamic equivalent friction coefficient and dynamic equivalent moment of inertia are calculated to establish a dynamic benchmark model. In the formal cutting segment, the cutting state shaft end mechanical output power of the X-axis and Y-axis servo drives is obtained, the non-cutting dissipation power is calculated by calling the dynamic benchmark model, and the pure cutting power component is obtained by subtracting the non-cutting dissipation power. The pure cutting power component is converted into a cutting resistance component, the actual cutting resistance vector angle is reconstructed, and the normalized phase deviation is obtained by comparing it with the theoretical resistance reference angle. The absolute phase deviation and the sliding window change rate of the absolute phase deviation are obtained from the normalized phase deviation. Based on the workpiece solid side normal to determine the micro-deflection polarity factor, real-time feed rate adjustment parameters are generated according to the normalized phase deviation, the absolute phase deviation, the sliding window change rate and the micro-deflection polarity factor, and real-time spindle speed adjustment parameters are generated in combination with the cutting load state, and respectively fed back to the trajectory interpolator and the spindle frequency converter to perform closed-loop intervention.

2. The CNC adaptive feed machining control method for glass cover plates according to claim 1, characterized in that, The steps for establishing the dynamic baseline model specifically include: The feed axis is determined to be in an idle state based on the instruction code of the CNC machining program and the axial coordinate of the tool spindle. When the absolute value of the feed axis command running speed in the idle state exceeds the effective speed threshold, the dynamic identification time window is activated. Within the dynamic identification time window, the parameters of the feedforward kinematic command and the mechanical output power of the unloaded shaft end are extracted synchronously to construct a discrete sampling data sample set; The observation matrix is ​​checked for full column rank, and the dynamic equivalent friction coefficient and the dynamic equivalent moment of inertia are solved by least squares regression.

3. The CNC adaptive feed machining control method for glass cover plates according to claim 1, characterized in that, The step of extracting the pure cutting power component specifically includes: The non-cutting power dissipation of the X-axis and Y-axis is obtained by multiplying the product of the dynamic equivalent friction coefficient and the absolute value of the feedforward command running speed, and then adding the product of the dynamic equivalent motion inertia, the feedforward command running acceleration, and the signed feedforward command running speed. The calculation result is obtained by subtracting the corresponding non-cutting dissipation power from the mechanical output power of the shaft end in the cutting state. When the calculation result is negative, the amplitude is limited to zero to obtain the pure cutting power component.

4. The CNC adaptive feed machining control method for glass cover plates according to claim 3, characterized in that, Following the step of extracting the pure cutting power component, the method further includes: Read the spindle speed, multiply it by the effective number of cutting edges, the number of grinding stages, or the equivalent periodic excitation parameters, and divide by 60 to obtain the periodic excitation reference frequency; The cutoff frequency is the product of the periodic excitation reference frequency and the preset ratio, and the first-order filter coefficient is calculated in combination with the inherent interpolation period of the trajectory interpolator. The first-order filter coefficients are used to perform a first-order discrete low-pass filter on the pure cutting power component.

5. The CNC adaptive feed machining control method for glass cover plates according to claim 4, characterized in that, The step of reconstructing the actual cutting resistance vector angle and comparing the actual cutting resistance vector angle with the theoretical trajectory tangential vector angle to obtain the normalized phase deviation specifically includes: When the absolute value of the feed axis command running speed is greater than or equal to the micro-speed determination threshold, the filtered pure cutting power component is divided by the corresponding feedforward command running speed, and the cutting resistance component is obtained after direction correction. When the absolute value of the feed axis command running speed is less than the micro-speed determination threshold, the friction torque and inertial torque are deducted by the torque current or equivalent disturbance torque, and the cutting resistance component is obtained by converting the transmission efficiency and the lead screw. The actual cutting resistance vector angle is synthesized by calling the four-quadrant arctangent function; The theoretical trajectory tangential vector angle plus the constant of pi is used as the theoretical resistance reference angle. The difference is calculated and normalized to obtain the normalized phase deviation.

6. The CNC adaptive feed machining control method for glass cover plates according to claim 5, characterized in that, The steps for determining the microscopic deflection polarity factor specifically include: The solid-side normal of the workpiece is determined based on the current internal and external contour attributes, programming direction, and tool radius compensation status. Based on the preset equivalent compliance matrix, the cutting resistance vector is converted into the tool tip micro displacement deflection vector; The micro-displacement deflection vector of the tool tip is projected onto the normal of the solid side of the workpiece. The value is assigned as 1 when the projected value is greater than the preset polarity recognition threshold, as -1 when the projected value is less than the opposite of the preset polarity recognition threshold, and as 0 when the absolute value of the projected value is not greater than the preset polarity recognition threshold.

7. The CNC adaptive feed machining control method for glass cover plates according to claim 6, characterized in that, The steps for generating real-time feed rate adjustment parameters specifically include: When the absolute phase deviation is greater than the absolute phase deviation tolerance threshold or the sliding window change rate is greater than the sliding window change rate tolerance threshold, and the micro-deflection polarity factor is assigned a value of 1, it is determined to be in an overcut state. Calculate the excess amount of the absolute phase deviation and the sliding window change rate respectively, and set the excess amount less than zero to zero; Calculate the feed rate attenuation coefficient based on the aforementioned deviation amount and the corresponding penalty weight; The real-time feed rate is obtained by subtracting the feed rate attenuation coefficient from the base rate command value and limiting it with the minimum safe rate threshold.

8. The CNC adaptive feed machining control method for glass cover plates according to claim 1, characterized in that, The step of generating real-time feed rate adjustment parameters further includes: When the absolute phase deviation is greater than the absolute phase deviation tolerance threshold or the sliding window change rate is greater than the sliding window change rate tolerance threshold, and the micro-deflection polarity factor is assigned a value of -1, it is determined to be in an undercut state. Extract the nominal feed increment vector and decompose it along the tangent of the theoretical trajectory and the normal of the workpiece solid side; The normal slow-release increment is generated according to the preset slow-release coefficient and superimposed with the nominal feed increment vector; Limits on single-cycle sustained-release increments and cumulative sustained-release increments; After the state is restored, a reverse compensation increment is generated based on the cumulative slow-release increment to smoothly revert to the nominal processing trajectory.

9. The CNC adaptive feed machining control method for glass cover plates according to claim 1, characterized in that, The steps for generating real-time spindle speed ratio adjustment parameters specifically include: Real-time reading of the spindle load rate, actual spindle speed, and the changing trend of the filtered pure cutting power component; When in the overcut state, reduce the real-time feed rate and maintain the reference spindle speed rate; When the spindle is in the undercut state and the spindle load rate does not exceed the preset spindle safe load threshold, the spindle speed ratio is adjusted between the preset upper limit ratio and the preset lower limit ratio according to the changing trend of the pure cutting power component.

10. A CNC adaptive feed machining control system for glass cover plates, comprising a CNC main control unit, a real-time communication bus, and drive execution nodes, characterized in that, Applicable to the method as described in any one of claims 1-9; The CNC main control unit is equipped with: The parameter identification module acquires the mechanical output power and feedforward kinematic commands of the unloaded shaft ends of the X-axis and Y-axis servo drives during the non-cutting idle stroke, calculates the dynamic equivalent friction coefficient and dynamic equivalent moment of inertia, and establishes a dynamic benchmark model. The power stripping module acquires the mechanical output power of the shaft end in the cutting state during the formal cutting segment, calls the dynamic benchmark model to calculate the non-cutting dissipation power, and subtracts the non-cutting dissipation power from the mechanical output power of the shaft end in the cutting state to extract the pure cutting power component. The deflection decoupling module calculates and reconstructs the actual cutting resistance vector angle by the pure cutting power component, compares it with the theoretical trajectory tangential vector angle to obtain the normalized phase deviation, calculates its sliding window change rate and determines the micro deflection polarity factor. The compensation control module generates real-time feed rate adjustment parameters based on the normalized phase deviation, sliding window change rate, and micro-deflection polarity factor, and generates real-time spindle speed adjustment parameters in conjunction with the cutting load state.

Images