Method and system for improving machining precision of precision screw machine based on adaptive control

Abstract

The application provides a precision screw machine tool machining precision improvement method and system based on adaptive control, relates to the technical field of machine tool machining, and comprises the following steps: acquiring geometric feature information of a screw to be machined and real-time running state information of a machine tool, extracting a thermal error, a vibration error and a tool offset component through multi-source error decoupling separation, determining each independent error component and a machining space distribution characteristic, reconstructing an original machining track into a compensation track embedded with an inverse compensation vector through space superposition mapping, establishing an inverse response tracing path to generate a prediction evolution sequence from a time domain variation extracted from the real-time running state information, and correcting a track offset in real time to obtain a dynamic compensation track, driving each motion shaft to perform machining and continuously updating the error component and the dynamic compensation track. The method effectively reduces the influence of multi-source errors and improves the machining precision and stability of precision screws.

Description

Technical Field

[0001] This invention relates to the field of machine tool processing technology, and in particular to a method and system for improving the machining accuracy of precision screws based on adaptive control. Background Technology

[0002] Precision screw machine tools are responsible for processing critical fasteners in high-end manufacturing fields such as consumer electronics, aerospace, and medical devices. Their machining accuracy directly determines the overall assembly quality and service reliability of the machine. With product miniaturization and continuous improvement in performance indicators, thread accuracy requirements have entered the micron or even sub-micron range, placing stringent demands on the dynamic characteristics and error suppression capabilities of machine tools. Current precision screw machining processes generally employ compensation strategies based on offline calibration and experience-based correction. This involves performing trial cuts and measurements on the machine tool before batch processing to obtain systematic positioning and geometric error data, which is then used to generate a fixed error compensation table and superimposed on the CNC code to achieve static correction of the machining trajectory. Parallel to this is the traditional PID feedback control loop built into the machine tool by the manufacturer. This loop adjusts the single-axis motion deviation in real time based on servo motor encoder feedback to maintain basic position tracking accuracy.

[0003] However, conventional technical procedures have limitations. The compensation model established through offline calibration has a fixed mapping relationship, and its compensation parameters are only valid under calibration conditions. When the machine tool undergoes long-term continuous operation, ambient temperature shifts, or cutting load fluctuations, thermal deformation and vibration characteristics will change significantly. The pre-written compensation table cannot detect such real-time changes and gradually becomes inaccurate, resulting in a significant drift in machining accuracy as the working time increases. Traditional feedback control and error compensation usually treat the overall machining deviation as a single whole for correction, without distinguishing and processing different physical sources such as thermal deformation, forced vibration, and tool wear. Various errors are coupled and superimposed, making it difficult for control actions to match the actual error generation mechanism. The compensation effect is severely limited, and in some cases, different error sources cancel each other out, greatly restricting the consistency and stability of precision screw machining accuracy. Summary of the Invention

[0004] This invention provides a method and system for improving the machining accuracy of precision screw machine tools based on adaptive control, which can solve the problems in the prior art.

[0005] A first aspect of the present invention provides a method for improving the machining accuracy of precision screw machine tools based on adaptive control, comprising: Obtain the geometric feature information of the screw to be processed and the real-time operating status information of the machine tool; Based on geometric feature information and real-time operating status information, the independent contribution components of each error source are identified through decoupling and separation processing of multi-source errors. The thermal error component, vibration error component and tool offset component are separated and extracted to obtain each independent error component, and the distribution characteristics of each independent error component in the machining space are determined. Based on each independent error component and its distribution characteristics, the original processing trajectory is reconstructed into a compensation trajectory through spatial superposition mapping relationship. The new spatial coordinates of each sampling point in the compensation trajectory are pre-embedded with the reverse compensation vector. The time-domain change is extracted from the real-time operating status information, a reverse response tracing path is established to determine the historical contribution ratio, a dissipation path and an enhancement path are constructed based on the historical contribution ratio to generate a predicted evolution sequence, and the trajectory offset is corrected in real time based on the predicted evolution sequence to obtain a dynamic compensation trajectory. The machine tool's motion axes are driven by dynamic compensation trajectories to perform machining actions. During the machining process, real-time operating status information is continuously collected and each independent error component and dynamic compensation trajectory are updated.

[0006] In one optional embodiment, identifying the independent contribution components of each error source through decoupling and separation processing of multi-source errors includes: The temperature field spatiotemporal evolution signal in the real-time operating status information is divided into regions according to the spatial positions of the machine tool bed, spindle box and tool post. The temperature gradient direction vector of each region is extracted. A coupling and transmission relationship between the temperature gradient direction vector of each region and the structural stiffness attenuation of each region is established. The local thermal expansion of materials caused by the temperature gradient is transformed into cross-regional cumulative displacement transmission. Through the cumulative displacement transmission, a nonlinear spatial distribution of thermally induced error components is formed in the machining space, and the thermally induced error components are obtained. Modal separation is performed on the vibration response signal in the real-time operating status information to identify the low-frequency modal components generated by the spindle rotation excitation and the high-frequency modal components generated by the tool cutting impact. The transmission attenuation path of the low-frequency modal components along the bed guide and the transmission amplification path of the high-frequency modal components along the tool post cantilever structure are established to obtain the vibration-induced error components. By correlating the screw pitch dimension change trend in the geometric feature information with the cutting force fluctuation amplitude in the tool wear monitoring signal, the asymmetric wear position of the tool edge is determined based on the periodic overload section of the tool caused by the pitch dimension change, and the tool offset component is obtained.

[0007] In one optional embodiment, determining the distribution characteristics of each independent error component in the processing space includes: The nonlinear spatial distribution of the thermal error component is projected and decomposed in the machining space along the machining trajectory direction to obtain the thermal displacement component tangential to the machining trajectory and the thermal displacement component perpendicular to the machining trajectory. The intersection of the low-frequency modal component along the bed guide rail and the high-frequency modal component along the tool post cantilever structure in the machining space is determined as the peak distribution area of ​​the vibration-induced error component. The periodic fluctuation amplitude and phase information of the vibration response signal are extracted from the peak distribution area and converted into the transient displacement deviation of the vibration-induced error component at each position in the machining space. Based on the asymmetric wear position of the tool edge, the spatial offset vector of the tool cutting edge relative to the preset tool position is reconstructed, and the spatial offset vector is decomposed along the machining trajectory direction to obtain the tangential component and normal component of the tool offset component; The transient displacement deviation of the thermally induced displacement component, the vibration-induced error component, and the tangential and normal components of the tool offset component are vector-superimposed at the same position in the machining space to obtain the distribution characteristics of each independent error component in the machining space.

[0008] In an optional embodiment, based on each independent error component and its distribution characteristics, the original processing trajectory is reconstructed into a compensation trajectory through a spatial superposition mapping relationship. The new spatial coordinates of each sampling point in the compensation trajectory are pre-embedded with a reverse compensation vector, including: The distribution characteristics of each independent error component are spatially projected onto the sampling point positions of the original processing trajectory. The projection weight coefficient is determined according to the spatial distance relationship between the distribution characteristics and the sampling point positions. The distribution characteristics are then mapped to the sampling point positions according to the projection weight coefficient to obtain the error projection vector at each sampling point position. The error projection vector is decomposed and traced back along multiple paths to obtain the reverse compensation vector at each sampling point location; A spatial superposition mapping relationship is established between the positions of each sampling point of the original processing trajectory and the reverse compensation vector. The spatial superposition mapping relationship is obtained by vector addition of the original spatial coordinates of each sampling point and the corresponding reverse compensation vector to obtain the new spatial coordinates of each sampling point after compensation. Extract the new spatial coordinates of each sampling point after compensation, and interpolate and connect the new spatial coordinates according to the temporal sequence of each sampling point in the original processing trajectory to generate a continuous compensated trajectory spatial curve, thus obtaining the compensated trajectory.

[0009] In one optional embodiment, the error projection vector is decomposed and traced back along multiple paths to obtain the reverse compensation vector at each sampling point location, including: For the error projection vector at each sampling point, the multi-path transmission characteristics of the error projection vector in the machining system are extracted. The multi-path transmission characteristics include the transmission component of the error projection vector to the machining point through the stiffness deformation of the machine tool structure, the transmission component to the machining point through the mechanical coupling between the tool and the workpiece contact interface, and the transmission component to the machining point through heat conduction from the heat source to the machining area. The transmission components are vector-superimposed at the machining point to form the actual machining deviation at that sampling point. Using the desired zero deviation state of the machining point position as a constraint, reverse tracing is performed along the machine tool structure stiffness deformation path, the mechanical coupling path of the tool and workpiece contact interface, and the heat conduction path from the heat source to the machining area to determine the tool position reverse adjustment component required to make the superposition of each transmission component at the machining point position zero. The reverse adjustment components of each tool position are vector synthesized at each sampling point to obtain the total tool trajectory adjustment vector at that sampling point as the reverse compensation vector.

[0010] In one optional embodiment, time-domain changes are extracted from real-time operating status information, a reverse response tracing path is established to determine the historical contribution ratio, a dissipation path and an enhancement path are constructed based on the historical contribution ratio to generate a predicted evolution sequence, and the trajectory offset is corrected in real time based on the predicted evolution sequence to obtain a dynamically compensated trajectory, including: Extract the time-domain changes of spindle speed, feed rate, and cutting load from real-time operating status information. Establish a reverse response tracing path for each independent error component. Traverse the reverse response tracing path from the current performance of each independent error component at the machining point to the source of the time-domain change that drives the current performance, and determine the historical contribution ratio of each time-domain change to each independent error component. The historical contribution ratio of each time-domain change is decomposed in the frequency domain. Based on the frequency domain decomposition results, the dissipation path and enhancement path of each independent error component in the future are constructed on the time axis. The first time distribution of natural error decay is determined along the dissipation path, and the second time distribution of continuous error amplification is determined along the enhancement path. The predicted evolution sequence of each independent error component is generated based on the first time distribution and the second time distribution. The dissipation and enhancement times in the predicted evolution sequence are mapped to the corresponding sampling point positions of the compensation trajectory. A zero compensation vector is applied to the sampling point positions corresponding to the dissipation times, and a reverse tracking compensation vector is applied to the sampling point positions corresponding to the enhancement times according to the amplification. The spatial coordinates of each sampling point are then updated to obtain the dynamic compensation trajectory.

[0011] In one optional embodiment, the historical contribution ratio of each time-domain change is decomposed in the frequency domain, and the dissipation path and enhancement path of each independent error component at future moments are constructed on the time axis based on the frequency domain decomposition results, including: The historical contribution ratios of each time-domain change are arranged in chronological order to form a contribution ratio time series. The contribution ratio time series is then decomposed in the frequency domain to extract the oscillation frequency and oscillation phase. The repetition period of each independent error component on the time axis is determined based on the oscillation frequency, and the phase start position of each independent error component within the repetition period is determined based on the oscillation phase. The repetition period and phase start position are periodically copied on the time axis after the current moment. The falling period and rising period of each independent error component transitioning from the phase peak to the phase valley are marked along the periodically copied time axis. The falling period is used as the dissipation path and the rising period is used as the enhancement path.

[0012] A second aspect of the present invention provides a precision screw machining accuracy improvement system based on adaptive control, comprising: The information acquisition unit is used to acquire the geometric feature information of the screw to be processed and the real-time operating status information of the machine tool; The error decoupling unit is used to identify the independent contribution components of each error source based on geometric feature information and real-time operating status information through decoupling and separation processing of multi-source errors. It separates and extracts thermal error components, vibration error components and tool offset components to obtain each independent error component and determines the distribution characteristics of each independent error component in the machining space. The trajectory compensation unit is used to reconstruct the original processing trajectory into a compensation trajectory based on each independent error component and its distribution characteristics through a spatial superposition mapping relationship. The new spatial coordinates of each sampling point in the compensation trajectory are pre-embedded with a reverse compensation vector. The dynamic correction unit is used to extract time-domain changes from real-time operating status information, establish a reverse response tracing path to determine the historical contribution ratio, construct dissipation path and enhancement path based on the historical contribution ratio to generate a predicted evolution sequence, and perform real-time correction of trajectory offset based on the predicted evolution sequence to obtain a dynamic compensation trajectory. The execution update unit is used to drive each motion axis of the machine tool to perform machining actions based on the dynamic compensation trajectory. During the machining process, it continuously collects real-time operating status information and updates each independent error component and the dynamic compensation trajectory.

[0013] A third aspect of the present invention provides an electronic device, comprising: processor; Memory used to store processor-executable instructions; The processor is configured to invoke instructions stored in the memory to execute the aforementioned method.

[0014] A fourth aspect of the present invention provides a computer-readable storage medium having stored thereon computer program instructions that, when executed by a processor, implement the aforementioned method.

[0015] In this embodiment of the invention, multi-source error decoupling and separation processing is used to clearly present the distribution characteristics of various error sources in the machining space. Error components no longer overlap, and their independent contributions to trajectory offset can be accurately evaluated and characterized. This avoids the risk of inaccurate compensation direction due to mutual cancellation or superposition of coupled errors, significantly improving the reliability of error identification and the targeting of compensation actions. Furthermore, the reconstruction of the original machining trajectory into a compensation trajectory is achieved through spatial superposition mapping. The new spatial coordinates of each sampling point are pre-embedded with the inverse compensation vector corresponding to the error vector at that point. The compensation amount directly acts on the trajectory geometry, rather than being indirectly canceled by servo gain adjustment, thus ensuring the tool... The deviation between the actual motion path and the theoretical path is systematically eliminated during the instruction generation stage. The point-by-point embedding compensation strategy effectively suppresses the accumulation of spatial nonlinear errors and has a good suppressive effect on local distortions in the machining of complex screw contours. A reverse response tracing path is established with time-domain changes as input, which correlates the current trajectory offset with the proportion of historical contributions, distinguishes between dissipative paths and enhancement paths, and generates a predictive evolution sequence. This can characterize the dynamic change trend of errors with the machining process, shifting the compensation action from ex-post correction to ex-ante prediction. During the machining process, real-time operating status information is continuously collected and each independent error component and dynamic compensation trajectory are updated synchronously, forming a closed-loop self-correction mechanism. The compensation strategy adaptively adjusts with changes in working conditions, maintaining accuracy stability in high-intensity continuous machining or multi-batch production, and reducing the frequency of repeated calibration and manual intervention. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating a method for improving the machining accuracy of precision screws on a precision machine tool based on adaptive control. Figure 2 This is a diagram showing the spatial coupling and overall migration distribution of multi-source errors; Figure 3 A logic flowchart for generating dynamic compensation trajectories; Figure 4 This is a comparison chart of dynamic look-ahead compensation and residual error suppression. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] The technical solution of the present invention will be described in detail below with reference to specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments.

[0019] Figure 1 This is a flowchart illustrating the method for improving the machining accuracy of precision screws on a precision machine tool based on adaptive control, according to an embodiment of the present invention.

[0020] Methods for improving the machining accuracy of precision screws on precision machine tools based on adaptive control include: Obtain the geometric feature information of the screw to be processed and the real-time operating status information of the machine tool; Based on geometric feature information and real-time operating status information, the independent contribution components of each error source are identified through decoupling and separation processing of multi-source errors. The thermal error component, vibration error component and tool offset component are separated and extracted to obtain each independent error component, and the distribution characteristics of each independent error component in the machining space are determined. Based on each independent error component and its distribution characteristics, the original processing trajectory is reconstructed into a compensation trajectory through spatial superposition mapping relationship. The new spatial coordinates of each sampling point in the compensation trajectory are pre-embedded with the reverse compensation vector. The time-domain change is extracted from the real-time operating status information, a reverse response tracing path is established to determine the historical contribution ratio, a dissipation path and an enhancement path are constructed based on the historical contribution ratio to generate a predicted evolution sequence, and the trajectory offset is corrected in real time based on the predicted evolution sequence to obtain a dynamic compensation trajectory. The machine tool's motion axes are driven by dynamic compensation trajectories to perform machining actions. During the machining process, real-time operating status information is continuously collected and each independent error component and dynamic compensation trajectory are updated.

[0021] In one optional embodiment, identifying the independent contribution components of each error source through decoupling and separation processing of multi-source errors includes: The temperature field spatiotemporal evolution signal in the real-time operating status information is divided into regions according to the spatial positions of the machine tool bed, spindle box and tool post. The temperature gradient direction vector of each region is extracted. A coupling and transmission relationship between the temperature gradient direction vector of each region and the structural stiffness attenuation of each region is established. The local thermal expansion of materials caused by the temperature gradient is transformed into cross-regional cumulative displacement transmission. Through the cumulative displacement transmission, a nonlinear spatial distribution of thermally induced error components is formed in the machining space, and the thermally induced error components are obtained. Modal separation is performed on the vibration response signal in the real-time operating status information to identify the low-frequency modal components generated by the spindle rotation excitation and the high-frequency modal components generated by the tool cutting impact. The transmission attenuation path of the low-frequency modal components along the bed guide and the transmission amplification path of the high-frequency modal components along the tool post cantilever structure are established to obtain the vibration-induced error components. By correlating the screw pitch dimension change trend in the geometric feature information with the cutting force fluctuation amplitude in the tool wear monitoring signal, the asymmetric wear position of the tool edge is determined based on the periodic overload section of the tool caused by the pitch dimension change, and the tool offset component is obtained.

[0022] In one specific implementation, when processing the spatiotemporal evolution signal of the temperature field in the real-time operating status information, the machine tool is divided into regions according to the spatial locations of three functional areas: the machine tool bed, the spindle box, and the tool post. Each region is equipped with a temperature sensor array to continuously collect temperature values ​​from various measuring points, thereby forming a temperature field dataset with both temporal and spatial dimensions. For each region, the spatial distribution of temperature values ​​at each measuring point within that region is extracted, and the temperature gradient direction vector is calculated. ,in Indicates the area code ( Corresponding to the bed frame, Corresponding spindle box, (Corresponding tool holder). Temperature gradient direction vector. It describes the direction of the most dramatic temperature changes in the region, reflecting the dominant path of heat flow.

[0023] Obtaining the temperature gradient direction vector of each region Next, it is necessary to establish the coupling and transmission relationship between this and the stiffness attenuation of the structure in each region. Since the elastic modulus of the machine tool's structural components changes under different temperature conditions, an increase in temperature will lead to a decrease in the local stiffness of the material, thus affecting the deformation response of the structure under external forces. Specifically, the local thermal expansion of the material in each region... Expressed as the temperature gradient magnitude and the regional thermal expansion coefficient and regional feature dimensions The product of, i.e. ,in For the first The coefficient of thermal expansion of the material in the region, For the first Characteristic dimensions of the region along the gradient direction. Local thermal expansion. The thermally induced error does not exist in isolation, but rather is transmitted to adjacent areas through the mechanical connection interfaces between different regions, forming a cumulative displacement transmission effect across regions. During this transmission process, the thermal expansion of the bed region first causes the bed guideway reference plane to shift, which in turn causes the spindle box and tool post to drift relative to the workpiece coordinate system. Mapping the superposition effect of the cumulative displacement transmission from each region onto the three-dimensional coordinate system of the machining space yields the nonlinear spatial distribution of the thermally induced error components in the machining space. Since the temperature field of each region continuously evolves over time, the thermally induced error components also exhibit dynamic characteristics that change over time, requiring updates to the temperature gradient direction vector in each control cycle. The cumulative displacement transfer is recalculated to ensure the real-time accuracy of the thermally induced error components.

[0024] The processing of vibration response signals begins with modal separation. Frequency domain analysis is performed on the acquired vibration response signals, decomposing them into multiple frequency components. The low-frequency modal components generated by spindle rotation excitation are related to the spindle speed and its harmonic frequencies, typically concentrated in the lower frequency band; the high-frequency modal components generated by tool cutting impact are closely related to cutting parameters such as the number of cutting teeth and feed rate, and have higher frequencies. Low-frequency modal components are extracted separately using bandpass filters. With high-frequency modal components This allows for effective separation of the two in the frequency domain. Low-frequency modal components When the vibration propagates within the machine tool bed guideway structure, due to the damping characteristics of the guideway, the vibration amplitude attenuates as the propagation distance increases. A transmission attenuation path for the low-frequency modal components along the bed guideway is established, and the attenuation coefficient is used... This describes the degree of amplitude attenuation per unit distance, thereby determining the displacement disturbance caused by low-frequency vibrations at different spatial locations on the machining point. High-frequency modal components. When the vibration propagates within the tool holder cantilever structure, due to the dynamic amplification effect of the cantilever structure, the vibration amplitude may tend to increase along the propagation path. This leads to the establishment of the high-frequency modal components along the propagation amplification path of the tool holder cantilever structure, utilizing the amplification factor... This describes the amplitude amplification factor of the cantilever end relative to the cantilever root. By combining the attenuation path analysis results of the low-frequency modal components and the amplification path analysis results of the high-frequency modal components, the influence of both on the tool tip position is superimposed in the machining space to obtain the vibration-induced error component. The vibration-induced error component exhibits periodic or quasi-periodic characteristics in the time domain, related to the spindle speed and cutting parameters, and its spatial distribution is closely related to the modal shape of the machine tool structure.

[0025] Determining the tool offset component requires correlating the screw pitch variation trend in the geometric feature information with the cutting force fluctuation amplitude in the tool wear monitoring signal. The screw pitch variation trend reflects the distribution of cutting load on the tool when machining screws with different pitch specifications. As the pitch increases, the contact length between the tool and the workpiece increases within each cutting stroke, and the peak cutting force experienced by the tool in a single cut rises accordingly, forming a periodic overload segment. By performing time-frequency analysis on the cutting force fluctuation amplitude signal, overload time segments where the cutting force exceeds the normal cutting range are identified. These overload segments are then time-aligned with the pitch variation cycle to determine at which cutting positions the tool repeatedly experiences excessive cutting force.

[0026] Because the cutting force on the tool in the overload zone is not perfectly symmetrical, the distribution of the cutting force on different sides of the tool edge varies, leading to asymmetric wear. By analyzing the directional distribution of the cutting force vector within the overload zone, the degree to which the cutting force is biased towards one side of the tool edge can be determined, thus pinpointing the specific location of asymmetric wear. Asymmetric wear of the tool edge causes the tool tip to shift laterally in a plane perpendicular to the cutting direction; this shift is the tool offset component. The magnitude of the tool offset component is positively correlated with the degree of wear; as the cumulative cutting time increases, the wear gradually accumulates, and the tool offset component also increases. In actual machining, by continuously monitoring the amplitude of cutting force fluctuations, the wear process can be tracked in real time, and the estimated value of the tool offset component can be dynamically updated, ensuring that the compensation calculation is always based on the latest tool condition information. By integrating the thermally induced error component, the vibration-induced error component, and the tool offset component, the decoupling and separation of multi-source errors can be achieved.

[0027] In one optional embodiment, determining the distribution characteristics of each independent error component in the processing space includes: The nonlinear spatial distribution of the thermal error component is projected and decomposed in the machining space along the machining trajectory direction to obtain the thermal displacement component tangential to the machining trajectory and the thermal displacement component perpendicular to the machining trajectory. The intersection of the low-frequency modal component along the bed guide rail and the high-frequency modal component along the tool post cantilever structure in the machining space is determined as the peak distribution area of ​​the vibration-induced error component. The periodic fluctuation amplitude and phase information of the vibration response signal are extracted from the peak distribution area and converted into the transient displacement deviation of the vibration-induced error component at each position in the machining space. Based on the asymmetric wear position of the tool edge, the spatial offset vector of the tool cutting edge relative to the preset tool position is reconstructed, and the spatial offset vector is decomposed along the machining trajectory direction to obtain the tangential component and normal component of the tool offset component; The transient displacement deviation of the thermally induced displacement component, the vibration-induced error component, and the tangential and normal components of the tool offset component are vector-superimposed at the same position in the machining space to obtain the distribution characteristics of each independent error component in the machining space.

[0028] In one specific implementation, when determining the distribution characteristics of each independent error component in the machining space, it is necessary to establish their respective spatial mapping relationships for the thermally induced error component, the vibration-induced error component, and the tool offset component, and then form a comprehensive distribution description through vector superposition.

[0029] Thermally induced error components typically exhibit a nonlinear spatial distribution in the machining space, resulting from the non-uniform expansion of machine tool components under thermal loads. When projecting and decomposing this nonlinear spatial distribution along the machining trajectory, a local coordinate system needs to be established at each sampling point on the machining trajectory. The tangent direction of the trajectory at that sampling point is taken as the tangential basis vector, and the direction perpendicular to the plane of the trajectory is taken as the normal basis vector. The projection of the spatial displacement vector of the thermally induced error component onto the tangential basis vector is the thermally induced displacement component along the tangential direction of the machining trajectory, denoted as […]. Characterizes the effect of thermal deformation on dimensional accuracy in the cutting feed direction; the projection perpendicular to the machining trajectory direction is denoted as... This characterizes the impact of thermal deformation on the accuracy of the machined contour shape. Because the helical trajectory of a screw thread has a continuously changing direction, the decomposition of the tangential and normal directions needs to be dynamically updated at each sampling point, rather than using a globally fixed coordinate system for static decomposition. In practice, the machining trajectory can be discretized into several sampling points, and the current tangential direction angle can be calculated at each sampling point. Then, the displacement vector of the thermally induced error components is projected onto the corresponding direction, thus obtaining the distribution along the entire machining trajectory. and sequence.

[0030] In determining the spatial distribution characteristics of vibration-induced error components, low-frequency modal components exhibit a transmission attenuation effect along the bed guideway, while high-frequency modal components exhibit a transmission amplification effect along the tool post cantilever structure. These two transmission paths intersect at a specific location in the machining space, where the vibration response superposition is most significant; therefore, this location is defined as the peak distribution region of the vibration-induced error components. Within this peak distribution region, after time-domain acquisition of the vibration response signal, the periodic fluctuation amplitude of the signal is extracted through frequency-domain analysis. With phase .in, Reflects the maximum transient displacement amplitude caused by vibration. This reflects the phase relationship between vibration and spindle rotation angle; both together determine the transient displacement deviation of the vibration-induced error component as it changes with time and spatial position during machining. Specifically, at any position in the processing space. At this point, transient displacement deviation The vibration amplitude and phase information at this location are comprehensively determined. The time-domain fluctuation signal needs to be mapped to the spatial coordinate domain by combining the spindle speed and feed rate, thus obtaining a spatial distribution description of the vibration-induced error component at various locations in the machining space. For thread machining, since there is a strict synchronization relationship between the spindle speed and feed motion, the phase information of the vibration signal has a particularly critical impact on the thread profile error. Therefore, it is necessary to ensure the time synchronization accuracy with the spindle encoder signal during phase extraction.

[0031] The spatial distribution characteristics of the tool offset component originate from the asymmetric wear of the tool edge. In actual machining, different parts of the tool edge experience differentiated wear due to uneven distribution of cutting loads, causing the actual spatial position of the cutting edge to shift relative to the preset tool position point. By measuring the wear morphology of the tool edge, the specific location of asymmetric wear can be identified, thereby reconstructing the spatial offset vector of the cutting edge relative to the preset tool position point. This vector describes the actual direction and amount of deviation of the cutting edge in three-dimensional space. Decompose along the machining trajectory direction, and similarly establish a local tangential and normal coordinate system at each sampling point to obtain the tangential component of the tool offset component. With normal components Tangential component The main factors affecting the pitch accuracy and axial dimension of the thread, as well as its normal component, are... It primarily affects the thread profile angle and radial dimension. In tool wear monitoring, the wear location can be dynamically identified by combining cutting force signals and acoustic emission signals, ensuring that the reconstruction accuracy of the offset vector is continuously updated as the machining process progresses.

[0032] After completing the spatial distribution analysis of the above three types of independent error components, the thermally induced displacement components (including...) and Transient displacement deviation of vibration-induced error components and the tangential component of the tool offset component With normal components Vector superposition is performed at the same location in the machining space. During vector superposition, it is essential to ensure that all components are performed within the same local coordinate system to avoid directional superposition errors caused by inconsistencies in coordinate systems. At each sampling point on the machining trajectory, the tangential contributions of each component are summed to obtain the comprehensive tangential error at that point. The combined normal error at that point is obtained by summing the normal contributions of each component. This results in the comprehensive distribution characteristics of each independent error component in the processing space. These comprehensive distribution characteristics are stored in the form of a sequence of sampling points, with each sampling point corresponding to a set of... The numerical values ​​fully describe the spatial distribution of the error field in the processing space. This distribution characteristic will serve as the basic input for subsequent compensation trajectory reconstruction, supporting the accurate embedding of the reverse compensation vector.

[0033] In engineering implementation, the density of sampling points on the machining trajectory needs to be set reasonably according to the thread pitch and machining accuracy requirements. For fine threads, the spacing between sampling points should be less than one-tenth of the pitch to ensure that the spatial resolution of the error distribution description meets the compensation accuracy requirements. The spatial distribution calculation of each error component can be initialized before machining starts and dynamically corrected during machining as real-time running status information is updated, thereby achieving adaptive tracking of error distribution characteristics.

[0034] like Figure 2 As shown, the figure illustrates the distribution characteristics of the combined error field after vector superposition of three independent error components—thermal-induced error, vibration-induced error, and tool offset—at the same location in the machining space. Different colored dashed lines depict the independent spatial trajectories of thermally induced displacement (orange), vibration transient deviation (purple), and tool offset (blue) along the sampling point sequence. A solid red line with a semi-transparent block highlights the combined error curve. The data evolution trend shows that the independent error components do not exhibit constant offsets but rather nonlinear fluctuations with the spatial location of the machining trajectory. Thermally induced error shows a gradual drift related to the accumulation of temperature gradients, vibration-induced error exhibits local peak regions due to modal convergence, and tool offset shows quasi-periodic perturbations related to the pitch period. The combined error curve is not a simple algebraic sum of the three amplitudes but rather the result of vector synthesis at each sampling point based on spatial projection weighting coefficients. Its amplitude fluctuations reflect the mutual reinforcement or cancellation effects of different error sources at specific spatial locations. This distribution feature, presented as a high-density sampling point sequence, fully describes the evolution of the error field along the entire helical machining trajectory. The spacing between sampling points is less than one-tenth of the pitch, ensuring that the spatial resolution meets the compensation requirements for precision thread machining. This distribution serves as the foundational input for subsequent reverse compensation vector embedding, and its accuracy directly determines the effectiveness of the compensation trajectory reconstruction. It is a crucial data support for achieving micrometer-level precision control in this invention.

[0035] In an optional embodiment, based on each independent error component and its distribution characteristics, the original processing trajectory is reconstructed into a compensation trajectory through a spatial superposition mapping relationship. The new spatial coordinates of each sampling point in the compensation trajectory are pre-embedded with a reverse compensation vector, including: The distribution characteristics of each independent error component are spatially projected onto the sampling point positions of the original processing trajectory. The projection weight coefficient is determined according to the spatial distance relationship between the distribution characteristics and the sampling point positions. The distribution characteristics are then mapped to the sampling point positions according to the projection weight coefficient to obtain the error projection vector at each sampling point position. The error projection vector is decomposed and traced back along multiple paths to obtain the reverse compensation vector at each sampling point location; A spatial superposition mapping relationship is established between the positions of each sampling point of the original processing trajectory and the reverse compensation vector. The spatial superposition mapping relationship is obtained by vector addition of the original spatial coordinates of each sampling point and the corresponding reverse compensation vector to obtain the new spatial coordinates of each sampling point after compensation. Extract the new spatial coordinates of each sampling point after compensation, and interpolate and connect the new spatial coordinates according to the temporal sequence of each sampling point in the original processing trajectory to generate a continuous compensated trajectory spatial curve, thus obtaining the compensated trajectory.

[0036] In one specific implementation, after obtaining the independent error components, such as the thermally induced error component, the vibration-induced error component, and the tool offset component, and their distribution characteristics in the machining space, it is necessary to accurately map this error information to each sampling point of the original machining trajectory, thereby constructing a compensation trajectory that can pre-compensate the errors. The original machining trajectory consists of a series of discrete sampling points, each carrying three-dimensional spatial coordinate information. The spacing between adjacent sampling points is adaptively set according to the geometric complexity of the screw thread. Sampling is appropriately densified in areas with large thread curvature to ensure the accuracy of trajectory reconstruction.

[0037] The distribution characteristics of each independent error component are not uniformly distributed throughout the machining space, but exhibit spatial non-uniformity closely related to the machine tool structure, heat source location, and vibration propagation path. To effectively transform these distribution characteristics into local error quantities at each sampling point, spatial projection processing of the distribution characteristics is required. Specifically, for each sampling point in the machining space, the spatial distance between the sampling point and each error distribution characteristic region is calculated, and the projection weight coefficient is determined based on this distance. The calculation of the projection weight coefficient adopts the distance attenuation principle: when the sampling point is close to a certain error distribution region, the influence weight of the distribution characteristics of that region on the sampling point is larger; when the distance is greater, the influence weight decreases with increasing distance. Let the th error component be... The sampling point and the first The Euclidean distance between the error distribution characteristic regions is Then the corresponding projection weight coefficient satisfy This normalization process ensures that the sum of the weights of all error distribution regions for the same sampling point is 1, thus guaranteeing the physical consistency of the projection process. By weighting and superimposing the various error distribution features according to the calculated projection weight coefficients, the first error distribution can be obtained. Error projection vector at each sampling point location This vector comprehensively reflects the spatial combined effect of three types of errors—thermal, vibration, and tool offset—at the sampling point.

[0038] After obtaining the error projection vectors at each sampling point, further multi-path propagation decomposition and reverse tracing are needed to obtain accurate reverse compensation vectors at each sampling point. The core idea of ​​multi-path propagation decomposition is that during the propagation of error from its source to the cutting point, it passes through multiple propagation paths in the machine tool's structural chain. The propagation characteristics (including structural stiffness, damping characteristics, and geometric constraints) differ along different paths, leading to variations in the contribution components of the same error source on different paths. A path propagation matrix is ​​then established for each propagation path. (in (For path numbering), project the error vector. The process involves decomposing the data along each path to obtain the decomposed components for each path. Reverse tracing involves mapping the path decomposed components from the cutting point to each sampling point in the reverse direction of each transmission path, eliminating the cumulative effects of structural deformation and displacement during the transmission process, thereby obtaining the compensation amount required at each sampling point. The reverse tracing results of each path are then combined to obtain the... The reverse compensation vector at each sampling point This vector is opposite in direction to the error projection vector and equal in magnitude to the compensation amount after the comprehensive error is corrected by the transmission path.

[0039] When establishing the spatial superposition mapping relationship between each sampling point of the original processing trajectory and the reverse compensation vector, let the first... The original spatial coordinates of each sampling point are: The corresponding reverse compensation vector is The new spatial coordinates after compensation It is obtained by vector addition, that is The physical meaning of this vector addition operation is: based on the original trajectory coordinates, the tool is pre-moved to a position opposite to the expected error direction, so that the actual error generated during machining pulls the tool back from the compensation position to the ideal machining position, thereby achieving active error cancellation. This spatial superposition mapping relationship is executed independently for each sampling point of the original machining trajectory, ensuring the full coverage and local accuracy of the compensation strategy in the entire machining trajectory space.

[0040] After calculating the new spatial coordinates for all sampling points, the discrete compensation coordinate points need to be reconnected into a continuous compensation trajectory spatial curve. Considering the strict requirements for trajectory continuity and smoothness in screw thread machining, the new spatial coordinate sequence is interpolated and connected according to the temporal order of the sampling points in the original machining trajectory. Cubic spline interpolation is used, which has the characteristic of continuous second derivative, ensuring a smooth curvature transition of the compensation trajectory between adjacent sampling points and avoiding impact on the machine tool's motion axes caused by trajectory bends introduced by interpolation. Cubic spline interpolation at adjacent sampling points... and A cubic polynomial curve segment is constructed between the curve segments. Each curve segment satisfies the boundary conditions of positional continuity, first derivative continuity, and second derivative continuity at the connection point, thereby ensuring the overall smoothness of the entire compensation trajectory space curve. For special positions such as the starting and ending points of the thread profile, a boundary condition setting method based on the slope of the clamping endpoints is adopted to ensure that the compensation trajectory matches the end constraints of the thread geometry.

[0041] After interpolation is completed, the generated compensation trajectory space curve is validated for rationality. This involves checking whether the offset of each compensation coordinate point exceeds the physical travel range of the machine tool's motion axes, and whether the curvature change between adjacent compensation points is within the response capability range of the machine tool's interpolation controller. If the validation reveals that the compensation amount in a local area is too large, the amplitude of the reverse compensation vector in that area is limited, and the interpolation connection for that area is re-executed to ensure the executability of the compensation trajectory at the engineering implementation level. The validated compensation trajectory serves as the reference trajectory for subsequent machining actions of each motion axis of the machine tool. The new spatial coordinates of each sampling point have a pre-embedded reverse compensation vector, which can effectively offset the combined effects of thermal errors, vibration errors, and tool offset errors on machining accuracy during processing, thereby significantly improving the machining accuracy and consistency of precision screw machine tools.

[0042] In one optional embodiment, the error projection vector is decomposed and traced back along multiple paths to obtain the reverse compensation vector at each sampling point location, including: For the error projection vector at each sampling point, the multi-path transmission characteristics of the error projection vector in the machining system are extracted. The multi-path transmission characteristics include the transmission component of the error projection vector to the machining point through the stiffness deformation of the machine tool structure, the transmission component to the machining point through the mechanical coupling between the tool and the workpiece contact interface, and the transmission component to the machining point through heat conduction from the heat source to the machining area. The transmission components are vector-superimposed at the machining point to form the actual machining deviation at that sampling point. Using the desired zero deviation state of the machining point position as a constraint, reverse tracing is performed along the machine tool structure stiffness deformation path, the mechanical coupling path of the tool and workpiece contact interface, and the heat conduction path from the heat source to the machining area to determine the tool position reverse adjustment component required to make the superposition of each transmission component at the machining point position zero. The reverse adjustment components of each tool position are vector synthesized at each sampling point to obtain the total tool trajectory adjustment vector at that sampling point as the reverse compensation vector.

[0043] In one specific implementation, after obtaining the error projection vectors at each sampling point, it is necessary to further decompose and reverse-trace these vectors along multiple error propagation paths existing in the machining system, thereby generating an accurate reverse compensation vector for each sampling point. Error propagation in the machine tool machining system is not a single path, but rather converges and superimposes at the machining point simultaneously through multiple physical mechanisms. Therefore, propagation models must be established for each type of propagation mechanism to achieve a complete description of the actual machining deviation.

[0044] For any i Each sampling point has an error projection vector. There are three main physical transmission paths in the machining system. The first path is the machine tool structural stiffness deformation path: the machine tool bed, column, spindle box and other structural components undergo elastic deformation under the action of cutting force and their own weight. This deformation is transmitted along the machine tool structural chain to the machining point, forming the structural stiffness deformation transmission component. The second path is the mechanical coupling path at the tool-workpiece contact interface: during the cutting process, the cutting force, friction, and contact stiffness between the tool and the workpiece jointly determine the mechanical response at the interface. The displacement deviation generated by this mechanical coupling effect is transmitted along the contact interface to the machining point, forming a mechanical coupling transmission component. The third path is the heat conduction path from the heat source to the machining area: cutting heat diffuses from the heat source (including the cutting zone, spindle bearings, guideway friction pairs, etc.) to the machining area through heat conduction, causing local thermal deformation and forming a heat conduction component. The transmission components of the three paths are vector-superimposed at the processing point location to constitute the actual processing deviation at that sampling point. Their relationship satisfies: ; The transmission components of each transmission path are represented by the path transmission matrix of the corresponding path. With error projection vector The path transfer matrix is ​​obtained through this process. It describes the process of the error starting from the error source and passing through the first... The amplitude attenuation, direction rotation, and phase change characteristics of the transmission path reaching the machining point are determined by the finite element stiffness analysis of the machine tool structure, the Hertzian contact mechanics model of the contact interface, and the numerical solution of the heat conduction equation. For structural stiffness deformation paths, The construction depends on the elastic modulus, moment of inertia, and constraint boundary conditions of each structural component of the machine tool; for mechanical coupling paths, The construction requires the introduction of cutting stiffness coefficient and contact stiffness parameters; for the heat conduction path, The construction is based on the thermal conductivity, specific heat capacity and thermal boundary conditions of each region.

[0045] After establishing the transmission models for each path, the expected zero-deviation state at the processing point location is used as a constraint to perform reverse tracing on each of the three transmission paths. The expected zero-deviation state requires that the vector superposition of the transmission components of the three paths at the processing point location is exactly zero, i.e.: ; To satisfy this constraint, a reverse tracing is performed along the machine tool structure stiffness deformation path to determine the reverse adjustment component that needs to be applied at the tool position. This ensures that, after the structural stiffness deformation path is transmitted, the compensation effect of the adjustment component at the processing point exactly cancels out the original structural deformation transmission component. The mathematical essence of reverse tracing is the path transfer matrix. Seek the generalized inverse, that is ,in for The pseudo-inverse matrix is ​​used to handle underdetermined or overdetermined conditions that may occur during the transfer process. Similarly, reverse tracing along the mechanical coupling path yields the reverse adjustment component of the tool position. By tracing back along the heat conduction path, the reverse adjustment component of the tool position is obtained. .

[0046] It is important to note that the coupling effects between different paths must be considered during the reverse tracing process. For example, thermal deformation caused by the heat conduction path can alter the local stiffness distribution of the machine tool structure, thus affecting the transmission characteristics of the structural stiffness deformation path. Similarly, changes in cutting force can affect the amount of cutting heat generated through mechanical coupling paths, thereby affecting the transmission components of the heat conduction path. To address this inter-path coupling, an iterative correction mechanism is introduced in the reverse tracing calculation: using the independent reverse tracing results of each path as initial values, the coupling terms in the transmission matrix of each path affected by other paths are gradually corrected until the changes in the reverse adjustment components of each path converge to a preset threshold. Within a certain range, meaning the difference in the Euclidean norm of each component in two consecutive iterations satisfies the convergence condition. This iterative correction process ensures the accuracy of the reverse tracing results under multi-path coupling conditions.

[0047] After completing the reverse tracing of the three paths and obtaining the corresponding reverse tool position adjustment components, , and Then, the three will be in the first Vector synthesis is performed at each sampling point location to obtain the total tool path adjustment vector at that sampling point location, i.e., the reverse compensation vector. : ; Vector synthesis must be performed in a unified machine tool coordinate system to ensure consistent directional representation of each adjustment component. Since reverse adjustment components from different paths may cancel each other out or superimpose and enhance each other in certain directions, the resulting vector synthesis... It can comprehensively reflect the optimal tool position adjustment required under the combined action of three transmission paths. The result... The three coordinate components correspond to , and That is, the reverse compensation vector in , , The axial component can be directly used for the first... Original spatial coordinates of each sampling point After correction, the new spatial coordinates are obtained after compensation. .

[0048] For screw thread machining, the tool moves along a helical trajectory during thread cutting, and the transmission characteristics of the machine tool structure stiffness deformation path will change along the tool path. The heat transfer components of the path change dynamically with the accumulation of the cutting stroke as the axial position changes. Therefore, the multi-path transfer decomposition and reverse tracing process needs to be executed one by one on all sampling points on the machining trajectory, and the transfer matrix of each path needs to be updated online according to the real-time collected running status information during the machining process to ensure that the reverse compensation vector always matches the current machining state, thereby achieving continuous improvement in screw machining accuracy.

[0049] like Figure 3 The diagram shown illustrates the logic flowchart for generating the dynamic compensation trajectory.

[0050] In one optional embodiment, time-domain changes are extracted from real-time operating status information, a reverse response tracing path is established to determine the historical contribution ratio, a dissipation path and an enhancement path are constructed based on the historical contribution ratio to generate a predicted evolution sequence, and the trajectory offset is corrected in real time based on the predicted evolution sequence to obtain a dynamically compensated trajectory, including: Extract the time-domain changes of spindle speed, feed rate, and cutting load from real-time operating status information. Establish a reverse response tracing path for each independent error component. Traverse the reverse response tracing path from the current performance of each independent error component at the machining point to the source of the time-domain change that drives the current performance, and determine the historical contribution ratio of each time-domain change to each independent error component. The historical contribution ratio of each time-domain change is decomposed in the frequency domain. Based on the frequency domain decomposition results, the dissipation path and enhancement path of each independent error component in the future are constructed on the time axis. The first time distribution of natural error decay is determined along the dissipation path, and the second time distribution of continuous error amplification is determined along the enhancement path. The predicted evolution sequence of each independent error component is generated based on the first time distribution and the second time distribution. The dissipation and enhancement times in the predicted evolution sequence are mapped to the corresponding sampling point positions of the compensation trajectory. A zero compensation vector is applied to the sampling point positions corresponding to the dissipation times, and a reverse tracking compensation vector is applied to the sampling point positions corresponding to the enhancement times according to the amplification. The spatial coordinates of each sampling point are then updated to obtain the dynamic compensation trajectory.

[0051] In one specific implementation, three types of time-domain variations are extracted from the real-time operating status information: the time-domain variation of spindle speed, the time-domain variation of feed rate, and the time-domain variation of cutting load. The time-domain variation of spindle speed reflects the periodic disturbance of the relative position between the tool and workpiece caused by speed fluctuations; the time-domain variation of feed rate records the non-uniformity of the feed axis during its movement, which will generate cumulative positional deviations on the machining trajectory; the time-domain variation of cutting load describes the dynamic fluctuations of cutting force over time, directly affecting the amplitude of tool offset and vibration response. These three types of time-domain variations are denoted as... , and ,in This represents a time variable. Inverse response tracing paths are established for each of the thermally induced error component, vibration-induced error component, and tool offset component. Following these paths, starting from the performance of each independent error component at the current machining point, the path is traced back step-by-step along the historical time axis to track the source of the temporal changes driving the current error performance.

[0052] The construction of the reverse response tracing path is based on the causal transmission relationship between each error component and the time-domain change. For the thermally induced error component, its current amplitude is closely related to the historical accumulation of cutting load and spindle speed fluctuations, because the conduction of heat in the machine tool structure has a significant time lag effect. For the vibration-induced error component, its frequency characteristics are directly coupled with the periodic changes in spindle speed, and sudden changes in feed rate will also excite the transient vibration response of the machine tool structure. For the tool offset component, the historical peak value of the cutting load determines the cumulative offset of the tool within the elastic deformation range. Along the reverse response tracing path, the historical contribution of each time-domain change to each independent error component is quantified as a contribution ratio, and the historical contribution ratio of each time-domain change corresponding to the thermally induced error component is denoted as... , , The historical contribution percentage of the vibration-induced error component is: , , The historical contribution percentage of the tool offset component is , , The sum of the proportions of each group is normalized to 1 to ensure the completeness of contribution allocation.

[0053] The historical contribution ratio of each time-domain variation is incorporated into the frequency domain decomposition process. , , Perform short-time Fourier transforms on each component to extract their energy distributions in different frequency bands, thus obtaining frequency domain representations. , , ,in This represents the frequency variable. Based on the frequency domain decomposition results, frequency components with a periodic decay trend and frequency components with a periodic amplification trend are identified among the independent error components. The periodic decay trend corresponds to a dissipation path, indicating that the error component will naturally decrease due to energy dissipation in the future; the periodic amplification trend corresponds to an enhancement path, indicating that the error component will continuously increase due to energy accumulation in the future. Along the dissipation path, the first-time distribution of the natural decay of the error is determined on the time axis, and the predicted amplitude of each independent error component in the first-time distribution is denoted as... subscript Indicates the error component category number, Indicates the first At each prediction time point, along the enhancement path, the distribution of the second time point on the time axis where the error continues to amplify is determined, and the prediction amplitude of each independent error component in the second time point distribution is denoted as [value]. .

[0054] Based on the distributions at the first and second time points, a predicted evolution sequence for each independent error component is generated. The predicted evolution sequence describes the amplitude trend of each independent error component at each future prediction time point on the time axis, and simultaneously marks each time point as either a dissipation time point or an enhancement time point. Dissipation times correspond to error amplitudes below a set threshold. At times when the error has negligible impact on machining accuracy, the error amplitude at the enhanced time exceeds [a certain value]. At specific moments, errors have a significant impact on machining accuracy, necessitating active compensation. The value is determined based on the dimensional tolerance requirements of screw processing, and is usually set to 10% to 20% of the processing tolerance zone width.

[0055] The dissipation and enhancement moments in the predicted evolution sequence are mapped to the corresponding sampling point positions of the compensation trajectory. Since each sampling point in the compensation trajectory corresponds to the processing time axis, an interpolation method is used to convert the dissipation and enhancement moments in the time domain into sampling point numbers in the trajectory space, thus achieving the mapping from the time domain prediction results to the spatial domain. For the sampling point positions corresponding to the dissipation moments, their reverse compensation vectors are set to zero, i.e., no additional compensation is applied, avoiding overcompensation for errors that have already tended to dissipate, which could introduce new biases. For the sampling point positions corresponding to the enhancement moments, the amplification magnitude of that moment in the predicted evolution sequence is considered. Calculate the reverse tracking compensation vector, denoted as the first... The reverse tracking compensation vector at each sampling point is Its direction is opposite to the direction of the prediction error vector at that sampling point, and its magnitude is the same as... Proportional, proportionality coefficient The calibration is performed based on the error component category and the characteristics of the processed material.

[0056] After applying the reverse tracking compensation vector, the spatial coordinates of each sampling point are updated. For the sampling point corresponding to the augmentation moment, the spatial coordinates of that sampling point in the original compensation trajectory are superimposed. The new spatial coordinates after dynamic compensation are obtained. For the sampling points corresponding to the dissipation time, the spatial coordinates remain unchanged from their original values ​​in the compensation trajectory. After the above processing, a complete trajectory containing dynamic compensation information is output, namely the dynamic compensation trajectory. The dynamic compensation trajectory is recalculated after each processing cycle with the update of real-time running status information, ensuring that the predicted evolution sequence always reflects the latest error evolution trend, thereby achieving continuous real-time correction of trajectory offset.

[0057] In actual machining, the vibration-induced error component caused by sudden changes in spindle speed often reaches its peak after several machining cycles. If compensation is based solely on the error measurement at the current moment, the compensation action will lag behind the appearance of the error peak. Through the aforementioned predictive evolution sequence mechanism, the reverse tracking compensation vector can be embedded into the corresponding sampling point before the error peak arrives, synchronizing the compensation action with the error peak on the time axis, significantly reducing residual errors caused by compensation lag. Furthermore, the identification of dissipation paths avoids applying unnecessary compensation during the natural dissipation phase of the error, reducing additional disturbances to the machine tool's motion axes caused by the compensation action and helping to maintain the overall stability of the machining process.

[0058] like Figure 4As shown, the actual suppression effect of the dynamic compensation strategy in the time domain is presented. By comparing the error amplitude changes of the "no compensation" and "dynamic compensation" trajectories, the effectiveness of the real-time correction mechanism based on the predicted evolution sequence is intuitively verified. The gray dashed line in the figure represents the original error fluctuation curve without any compensation. Its amplitude shows a clear periodic increasing trend in the sampling time intervals of approximately 5-15 and 25-35, corresponding to the "enhancement path" period where the error continues to amplify. The green solid line represents the residual error curve after dynamic compensation. It can be observed that in the enhancement path interval (the amplitude rising segment in the figure), the dynamic compensation curve actively outputs a reverse tracking compensation vector proportional to the predicted amplification amplitude, significantly reducing the error peak value and suppressing the maximum deviation from approximately 2.8 μm to below 1.0 μm. In the dissipation path interval (the error natural decay segment), the compensation amount adaptively returns to zero, and the residual error tends to coincide with the no compensation curve, effectively avoiding the additional disturbances introduced by overcompensation.

[0059] In one optional embodiment, the historical contribution ratio of each time-domain change is decomposed in the frequency domain, and the dissipation path and enhancement path of each independent error component at future moments are constructed on the time axis based on the frequency domain decomposition results, including: The historical contribution ratios of each time-domain change are arranged in chronological order to form a contribution ratio time series. The contribution ratio time series is then decomposed in the frequency domain to extract the oscillation frequency and oscillation phase. The repetition period of each independent error component on the time axis is determined based on the oscillation frequency, and the phase start position of each independent error component within the repetition period is determined based on the oscillation phase. The repetition period and phase start position are periodically copied on the time axis after the current moment. The falling period and rising period of each independent error component transitioning from the phase peak to the phase valley are marked along the periodically copied time axis. The falling period is used as the dissipation path and the rising period is used as the enhancement path.

[0060] In one specific implementation, after obtaining the historical contribution percentage of each time-domain change to each independent error component, it is necessary to further explore the inherent patterns of the evolution of these contribution percentages over time to support the prediction of error behavior at future moments. To this end, the historical contribution percentages of each time-domain change are arranged sequentially in chronological order, forming their respective corresponding contribution percentage time series. The historical contribution percentage of the spindle speed time-domain change corresponding to the thermally induced error component is used as an example. For example, arranging them in chronological order of sampling time within a historical observation window yields a time-series signal reflecting the fluctuations in the contribution percentage over time. Similarly, for... , , , , , , , Perform the same time series arrangement operation on each component to form a complete time series set of contribution proportions for each error component and each time domain change.

[0061] The time series for each contribution percentage are decomposed in the frequency domain. Specifically, a Discrete Fourier Transform is applied to each contribution percentage time series to transform the time-domain sequence into the frequency domain, thereby obtaining the energy distribution of the time series at different frequency components. In the frequency domain representation, the frequency component with the most concentrated energy is the dominant oscillation frequency. Let the i-th... The contribution ratio of a certain time-domain change to a type of error component corresponds to the oscillation frequency of the time series. The unit is Hertz (Hz), and its physical meaning is the dominant frequency of periodic fluctuations in which this contribution occurs within a historical period. Simultaneously, the dominant oscillation frequency is extracted from the frequency domain decomposition results. The corresponding phase angle is denoted as the oscillation phase. The unit is radians. The dominant oscillation frequency and oscillation phase together describe the dominant oscillation characteristics of the contribution proportion time series, and are the core basis for subsequent construction of dissipation paths and enhancement paths.

[0062] Based on the oscillation frequency Determine the repetition period of each independent error component on the time axis. Repetition Period With oscillation frequency Between satisfy ,in The unit is seconds. This repetition period characterizes the time required for the corresponding error component to complete one full oscillation cycle. For different error component categories... Its repetition period The differences may exist, reflecting the different evolution rhythms of thermally induced errors, vibration-induced errors, and tool offset errors over time. For example, the repetition period of the thermally induced error component is usually long, corresponding to the slow process of establishing and dissipating machine tool thermal equilibrium; the repetition period of the vibration-induced error component is relatively short, closely related to the excitation frequency of spindle rotation and feed motion; and the repetition period of the tool offset component depends on the rhythmic characteristics of tool wear accumulation and recovery.

[0063] According to the oscillation phase Determine the phase start position of each independent error component within the repetition period. The phase start position is defined as the position at the current time. At this point, the phase angle position of the timing oscillation waveform corresponds to the contribution percentage of the corresponding error component. This is determined by the oscillation phase. With the current moment By comparing the corresponding phase angles, it can be determined which stage of the error component's repetition cycle the current moment falls into, i.e., whether the current moment is in the rising phase, near the peak, falling phase, or near the trough. This information is crucial for accurately predicting the evolution trend of the error component in future moments, avoiding the prediction phase shift problem caused by ignoring the initial phase conditions.

[0064] Repeating cycle The phase start position is periodically copied along the timeline after the current moment. Specifically, it is operated as follows: using the current moment... Starting from the repeating cycle By replicating the complete waveform structure of each repetition period sequentially along the future time axis at intervals, a continuously extending periodic oscillation prediction curve is formed in the prediction time domain. The dominant oscillation frequency must be maintained during periodic replication. With oscillation phase Consistency is ensured to guarantee a smooth transition between the predicted curve and historical observation time series at the current moment, without phase jumps or amplitude abrupt changes. In practice, the length of the prediction time domain is determined comprehensively based on the total duration of the processing task and the real-time update frequency, typically covering several repetition cycles. This ensures that the predicted sequence can support the dynamic compensation requirements throughout the entire processing.

[0065] Along the time axis of periodic replication, the phase intervals of the predicted oscillation curves for each independent error component are marked. In each repetition cycle... Within the predicted oscillation curve, there will be a declining period transitioning from the phase peak to the phase trough, and a rising period transitioning from the phase trough to the phase peak. The declining period corresponds to the process of the error component's contribution decreasing from a high level to a low level. During this process, the actual impact of the error component tends to weaken, therefore the declining period is defined as the dissipation path. The rising period corresponds to the process of the error component's contribution increasing from a low level to a high level. During this process, the actual impact of the error component tends to strengthen, therefore the rising period is defined as the strengthening path. The boundary points between the dissipation path and the strengthening path are the peak and trough times of the oscillation curve, which can be accurately determined by taking the first derivative of the predicted oscillation curve and setting it to zero.

[0066] When labeling dissipation paths and enhancement paths, it is also necessary to consider the error magnitude threshold. The effective range of the path is filtered. When the amplitude of the predicted oscillation curve is higher than... Only when the amplitude is below a certain value does the path marker for the corresponding time period have compensatory significance and be included in the effective dissipation path or effective enhancement path; When the error component in a given period is considered negligible, the corresponding path segment is not included in the subsequent prediction evolution sequence construction. This screening mechanism effectively avoids overcompensation for minor error fluctuations, reduces the computational cost of dynamic compensation trajectories, and ensures that compensation resources are concentrated on periods with significant error amplitudes.

[0067] Through the aforementioned frequency domain decomposition and path marking process, a clear dissipation path and enhancement path structure is ultimately established for each type of independent error component on the future time axis. These path structures are directly linked to the generation of the predicted evolution sequence: at the prediction time corresponding to the enhancement path, the predicted amplitude of the error component shows an upward trend, requiring a larger reverse compensation to be applied in advance in the dynamic compensation trajectory; at the prediction time corresponding to the dissipation path, the predicted amplitude of the error component shows a downward trend, and the corresponding reverse compensation amplitude can be reduced accordingly. This path construction method based on frequency domain decomposition enables the dynamic compensation trajectory to proactively respond to the periodic evolution law of the error components, rather than relying solely on the error measurement value at the current moment for lag compensation, thereby significantly improving the trajectory control accuracy and error suppression timeliness in the precision screw machine tool processing process.

[0068] A second aspect of the present invention provides a precision screw machining accuracy improvement system based on adaptive control, comprising: The information acquisition unit is used to acquire the geometric feature information of the screw to be processed and the real-time operating status information of the machine tool; The error decoupling unit is used to identify the independent contribution components of each error source based on geometric feature information and real-time operating status information through decoupling and separation processing of multi-source errors. It separates and extracts thermal error components, vibration error components and tool offset components to obtain each independent error component and determines the distribution characteristics of each independent error component in the machining space. The trajectory compensation unit is used to reconstruct the original processing trajectory into a compensation trajectory based on each independent error component and its distribution characteristics through a spatial superposition mapping relationship. The new spatial coordinates of each sampling point in the compensation trajectory are pre-embedded with a reverse compensation vector. The dynamic correction unit is used to extract time-domain changes from real-time operating status information, establish a reverse response tracing path to determine the historical contribution ratio, construct dissipation path and enhancement path based on the historical contribution ratio to generate a predicted evolution sequence, and perform real-time correction of trajectory offset based on the predicted evolution sequence to obtain a dynamic compensation trajectory. The execution update unit is used to drive each motion axis of the machine tool to perform machining actions based on the dynamic compensation trajectory. During the machining process, it continuously collects real-time operating status information and updates each independent error component and the dynamic compensation trajectory.

[0069] A third aspect of the present invention provides an electronic device, comprising: processor; Memory used to store processor-executable instructions; The processor is configured to invoke instructions stored in the memory to execute the aforementioned method.

[0070] A fourth aspect of the present invention provides a computer-readable storage medium having stored thereon computer program instructions that, when executed by a processor, implement the aforementioned method.

[0071] This invention can be a method, apparatus, system, and / or computer program product. The computer program product may include a computer-readable storage medium having computer-readable program instructions loaded thereon for performing various aspects of the invention.

[0072] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for improving the machining accuracy of precision screws on a precision machine tool based on adaptive control, characterized in that, include: Obtain the geometric feature information of the screw to be processed and the real-time operating status information of the machine tool; Based on geometric feature information and real-time operating status information, the independent contribution components of each error source are identified through decoupling and separation processing of multi-source errors. The thermal error component, vibration error component and tool offset component are separated and extracted to obtain each independent error component, and the distribution characteristics of each independent error component in the machining space are determined. Based on each independent error component and its distribution characteristics, the original processing trajectory is reconstructed into a compensation trajectory through spatial superposition mapping relationship. The new spatial coordinates of each sampling point in the compensation trajectory are pre-embedded with the reverse compensation vector. The time-domain change is extracted from the real-time operating status information, a reverse response tracing path is established to determine the historical contribution ratio, a dissipation path and an enhancement path are constructed based on the historical contribution ratio to generate a predicted evolution sequence, and the trajectory offset is corrected in real time based on the predicted evolution sequence to obtain a dynamic compensation trajectory. The machine tool's motion axes are driven by dynamic compensation trajectories to perform machining actions. During the machining process, real-time operating status information is continuously collected and each independent error component and dynamic compensation trajectory are updated.

2. The method according to claim 1, characterized in that, The independent contribution components of each error source are identified through decoupling and separation of multi-source errors, including: The temperature field spatiotemporal evolution signal in the real-time operating status information is divided into regions according to the spatial positions of the machine tool bed, spindle box and tool post. The temperature gradient direction vector of each region is extracted. A coupling and transmission relationship between the temperature gradient direction vector of each region and the structural stiffness attenuation of each region is established. The local thermal expansion of materials caused by the temperature gradient is transformed into cross-regional cumulative displacement transmission. Through the cumulative displacement transmission, a nonlinear spatial distribution of thermally induced error components is formed in the machining space, and the thermally induced error components are obtained. Modal separation is performed on the vibration response signal in the real-time operating status information to identify the low-frequency modal components generated by the spindle rotation excitation and the high-frequency modal components generated by the tool cutting impact. The transmission attenuation path of the low-frequency modal components along the bed guide and the transmission amplification path of the high-frequency modal components along the tool post cantilever structure are established to obtain the vibration-induced error components. By correlating the screw pitch dimension change trend in the geometric feature information with the cutting force fluctuation amplitude in the tool wear monitoring signal, the asymmetric wear position of the tool edge is determined based on the periodic overload section of the tool caused by the pitch dimension change, and the tool offset component is obtained.

3. The method according to claim 2, characterized in that, Determining the distribution characteristics of each independent error component in the machining space includes: The nonlinear spatial distribution of the thermal error component is projected and decomposed in the machining space along the machining trajectory direction to obtain the thermal displacement component tangential to the machining trajectory and the thermal displacement component perpendicular to the machining trajectory. The intersection of the low-frequency modal component along the bed guide rail and the high-frequency modal component along the tool post cantilever structure in the machining space is determined as the peak distribution area of ​​the vibration-induced error component. The periodic fluctuation amplitude and phase information of the vibration response signal are extracted from the peak distribution area and converted into the transient displacement deviation of the vibration-induced error component at each position in the machining space. Based on the asymmetric wear position of the tool edge, the spatial offset vector of the tool cutting edge relative to the preset tool position is reconstructed, and the spatial offset vector is decomposed along the machining trajectory direction to obtain the tangential component and normal component of the tool offset component; The transient displacement deviation of the thermally induced displacement component, the vibration-induced error component, and the tangential and normal components of the tool offset component are vector-superimposed at the same position in the machining space to obtain the distribution characteristics of each independent error component in the machining space.

4. The method according to claim 1, characterized in that, Based on the independent error components and their distribution characteristics, the original processing trajectory is reconstructed into a compensation trajectory through spatial superposition mapping. The new spatial coordinates of each sampling point in the compensation trajectory are pre-embedded with a reverse compensation vector, including: The distribution characteristics of each independent error component are spatially projected onto the sampling point positions of the original processing trajectory. The projection weight coefficient is determined according to the spatial distance relationship between the distribution characteristics and the sampling point positions. The distribution characteristics are then mapped to the sampling point positions according to the projection weight coefficient to obtain the error projection vector at each sampling point position. The error projection vector is decomposed and traced back along multiple paths to obtain the reverse compensation vector at each sampling point location; A spatial superposition mapping relationship is established between the positions of each sampling point of the original processing trajectory and the reverse compensation vector. The spatial superposition mapping relationship is obtained by vector addition of the original spatial coordinates of each sampling point and the corresponding reverse compensation vector to obtain the new spatial coordinates of each sampling point after compensation. Extract the new spatial coordinates of each sampling point after compensation, and interpolate and connect the new spatial coordinates according to the temporal sequence of each sampling point in the original processing trajectory to generate a continuous compensated trajectory spatial curve, thus obtaining the compensated trajectory.

5. The method according to claim 4, characterized in that, The error projection vector is decomposed and traced back along multiple paths to obtain the reverse compensation vector at each sampling point location, including: For the error projection vector at each sampling point, the multi-path transmission characteristics of the error projection vector in the machining system are extracted. The multi-path transmission characteristics include the transmission component of the error projection vector to the machining point through the stiffness deformation of the machine tool structure, the transmission component to the machining point through the mechanical coupling of the tool and workpiece contact interface, and the transmission component to the machining point through heat conduction from the heat source to the machining area. The transmission components are vector-superimposed at the machining point to form the actual machining deviation at that sampling point. Using the desired zero deviation state of the machining point position as a constraint, reverse tracing is performed along the machine tool structure stiffness deformation path, the mechanical coupling path of the tool and workpiece contact interface, and the heat conduction path from the heat source to the machining area to determine the tool position reverse adjustment component required to make the superposition of each transmission component at the machining point position zero. The reverse adjustment components of each tool position are vector synthesized at each sampling point to obtain the total tool trajectory adjustment vector at that sampling point as the reverse compensation vector.

6. The method according to claim 1, characterized in that, The time-domain changes are extracted from real-time operational status information. A reverse response tracing path is established to determine the historical contribution ratio. Based on the historical contribution ratio, dissipation paths and enhancement paths are constructed to generate a predicted evolution sequence. The trajectory offset is corrected in real time based on the predicted evolution sequence to obtain a dynamically compensated trajectory, including: Extract the time-domain changes of spindle speed, feed rate, and cutting load from real-time operating status information. Establish a reverse response tracing path for each independent error component. Traverse the reverse response tracing path from the current performance of each independent error component at the machining point to the source of the time-domain change that drives the current performance, and determine the historical contribution ratio of each time-domain change to each independent error component. The historical contribution ratio of each time-domain change is decomposed in the frequency domain. Based on the frequency domain decomposition results, the dissipation path and enhancement path of each independent error component in the future are constructed on the time axis. The first time distribution of natural error decay is determined along the dissipation path, and the second time distribution of continuous error amplification is determined along the enhancement path. The predicted evolution sequence of each independent error component is generated based on the first time distribution and the second time distribution. The dissipation and enhancement times in the predicted evolution sequence are mapped to the corresponding sampling point positions of the compensation trajectory. A zero compensation vector is applied to the sampling point positions corresponding to the dissipation times, and a reverse tracking compensation vector is applied to the sampling point positions corresponding to the enhancement times according to the amplification. The spatial coordinates of each sampling point are then updated to obtain the dynamic compensation trajectory.

7. The method according to claim 6, characterized in that, The historical contribution ratio of each time-domain change is decomposed in the frequency domain. Based on the frequency domain decomposition results, the dissipation path and enhancement path of each independent error component at future times are constructed on the time axis, including: The historical contribution ratios of each time-domain change are arranged in chronological order to form a contribution ratio time series. The contribution ratio time series is then decomposed in the frequency domain to extract the oscillation frequency and oscillation phase. The repetition period of each independent error component on the time axis is determined based on the oscillation frequency, and the phase start position of each independent error component within the repetition period is determined based on the oscillation phase. The repetition period and phase start position are periodically copied on the time axis after the current moment. The falling period and rising period of each independent error component transitioning from the phase peak to the phase valley are marked along the periodically copied time axis. The falling period is used as the dissipation path and the rising period is used as the enhancement path.

8. A precision screw machining accuracy improvement system based on adaptive control, used to implement the method as described in any one of claims 1-7, characterized in that, include: The information acquisition unit is used to acquire the geometric feature information of the screw to be processed and the real-time operating status information of the machine tool; The error decoupling unit is used to identify the independent contribution components of each error source based on geometric feature information and real-time operating status information through decoupling and separation processing of multi-source errors. It separates and extracts thermal error components, vibration error components and tool offset components to obtain each independent error component and determines the distribution characteristics of each independent error component in the machining space. The trajectory compensation unit is used to reconstruct the original processing trajectory into a compensation trajectory based on each independent error component and its distribution characteristics through a spatial superposition mapping relationship. The new spatial coordinates of each sampling point in the compensation trajectory are pre-embedded with a reverse compensation vector. The dynamic correction unit is used to extract time-domain changes from real-time operating status information, establish a reverse response tracing path to determine the historical contribution ratio, construct dissipation path and enhancement path based on the historical contribution ratio to generate a predicted evolution sequence, and perform real-time correction of trajectory offset based on the predicted evolution sequence to obtain a dynamic compensation trajectory. The execution update unit is used to drive each motion axis of the machine tool to perform machining actions based on the dynamic compensation trajectory. During the machining process, it continuously collects real-time operating status information and updates each independent error component and the dynamic compensation trajectory.

9. An electronic device, characterized in that, include: processor; Memory used to store processor-executable instructions; The processor is configured to invoke instructions stored in the memory to execute the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having computer program instructions stored thereon, characterized in that, When the computer program instructions are executed by the processor, they implement the method described in any one of claims 1 to 7.

Images