Wood cutting load self-adaptive CNC feeding speed dynamic adjusting method

Abstract

The invention discloses a wood cutting load self-adaptive CNC feeding speed dynamic adjusting method. The method comprises the steps that a wood digital twin voxel model containing physical attributes is constructed; virtual cutting simulation is carried out in the look-ahead time window according to the tool path, a to-be-removed voxel set is extracted, and theoretical cutting power is calculated and predicted in combination with the texture direction influence; a spindle motor signal is collected, and real-time cutting net power is obtained after no load is filtered out; comparing the real-time power with the predicted power, and calculating geometric and physical components in a normalized hardness factor decoupling load; and generating a geometric constraint adjusting quantity according to the predicted power and generating a physical self-adaptive adjusting quantity according to the hardness factor, and jointly synthesizing a target feeding speed instruction to dynamically control the machine tool. According to the method, prospective avoidance of geometric loads and quick response to material mutation are achieved, the problems that adjustment is lagged and load sources cannot be distinguished in a traditional method are solved, and the efficiency, quality and safety of CNC machining of wood are improved.

Description

Technical Field

[0001] This invention relates to the field of CNC machining technology, and in particular to a method for dynamically adjusting the CNC feed rate to adapt to wood cutting load. Background Technology

[0002] Numerical control (CNC) machining technology, as the cornerstone of modern manufacturing, aims to achieve high-efficiency, high-precision automated cutting. During machining, feed rate is a key process parameter determining machining efficiency, surface quality, and tool life. Traditional CNC machining typically uses a constant feed rate preset based on G-code. This value is often conservatively set based on experience to avoid the worst-case scenario, resulting in low machining efficiency in areas with low cutting loads. To address this issue, adaptive control machining (ACM) technology emerged. Early adaptive control systems relied primarily on real-time feedback, passively reducing the feed rate when the cutting load exceeded a preset threshold by monitoring the spindle motor's current or power. While this reactive control mode avoids overload to some extent, its response is lag-dependent. Sudden changes in load, especially in high-speed cutting, can still damage the tool or workpiece before the system adjusts.

[0003] However, existing adaptive control technologies exhibit significant limitations when applied to natural heterogeneous materials such as wood. Firstly, wood, as an anisotropic biomaterial, has extremely uneven distribution of its internal physical properties (such as density and hardness), and commonly exhibits complex characteristics such as grain direction, knots, decay, or cavities. These factors all lead to drastic fluctuations in cutting resistance. Traditional feedback control systems, relying solely on a single load signal, cannot effectively distinguish whether changes in cutting load stem from alterations in cutting geometry parameters (such as cutting depth and width) or from abrupt changes in local material physical properties, thus hindering precise optimization. Secondly, the cutting force in wood is closely related to the tool's movement direction relative to the wood grain direction. The energy consumption difference between cutting with and across the grain is significant, but existing technologies typically ignore the prior influence of this anisotropy on the cutting load.

[0004] In conclusion, developing a dynamic feed rate adjustment method that can predict geometric cutting loads, decouple and identify changes in material physical properties in real time, and comprehensively consider the influence of material anisotropy is of great significance for improving the efficiency, quality and intelligence level of wood CNC machining. Summary of the Invention

[0005] The purpose of this section is to outline some aspects of embodiments of the present invention and to briefly describe some preferred embodiments. Simplifications or omissions may be made in this section, as well as in the abstract and title of this application, to avoid obscuring the purpose of these documents; however, such simplifications or omissions should not be construed as limiting the scope of the invention.

[0006] In view of the aforementioned existing problems, this invention is proposed. Therefore, this invention provides a CNC feed rate dynamic adjustment method that adapts to wood cutting load, to solve the problems mentioned in the background art.

[0007] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a method for dynamically adjusting the CNC feed rate to adapt to wood cutting load, comprising: A digital twin model of the wood to be processed is constructed in the CNC system, and the wood entity is discretized into a three-dimensional voxel matrix containing physical properties. Within a preset look-ahead time window before the actual cutting occurs, a sweep volume of a virtual tool is generated according to the tool path instructions of the CNC machining program, and the Boolean intersection operation between the sweep volume and the three-dimensional voxel matrix is ​​performed to extract the set of voxels to be removed. Based on the total geometric volume of the set of voxels to be removed, and considering the effect of texture direction on the correction of cutting resistance, the predicted theoretical cutting power within the time window is calculated. During actual machining, the real-time drive signal of the spindle motor is collected synchronously, and the real-time net cutting power is obtained after filtering out the no-load loss. The real-time net cutting power is compared with the predicted theoretical cutting power to calculate the normalized hardness factor that reflects the properties of wood material, thereby decoupling the geometric cutting component and the physical material component in the load. The geometric constraint adjustment amount generated by the predicted theoretical cutting power and the physical adaptive adjustment amount generated by the normalized hardness factor are used to synthesize the target feed speed command and dynamically control the CNC machine tool.

[0008] In a preferred embodiment of the adaptive CNC feed rate dynamic adjustment method for wood cutting load described in this invention, each voxel unit in the three-dimensional voxel matrix is ​​assigned structured attribute data, which includes: The location status identifier representing the presence or absence of the voxel, the unit vector representing the direction of wood grain growth at the location of the voxel, and the preset reference value representing the hardness of the base material of the voxel. The three-dimensional voxel matrix is ​​stored using a spatial index structure to support fast retrieval.

[0009] As a preferred embodiment of the CNC feed rate dynamic adjustment method for adaptive wood cutting load described in this invention, the Boolean intersection operation includes: A continuous spatial envelope is generated based on the tool motion trajectory and tool geometry within the current interpolation cycle. A spatial index structure is used to retrieve voxels located within the spatial envelope and whose position status is identified as existing. These voxels are marked as the cutting objects at the current moment, and their total volume is calculated to obtain the instantaneous material removal rate.

[0010] As a preferred embodiment of the adaptive CNC feed rate dynamic adjustment method for wood cutting load described in this invention, the calculation method for the effect of the grain direction on the correction of cutting resistance includes: For each voxel in the set of voxels to be removed, the current tool feed synthesis speed vector and the texture growth direction unit vector recorded by the voxel are obtained respectively. Calculate the spatial angle between the feed synthesis velocity vector and the texture growth direction unit vector; The anisotropic cutting energy correction coefficient is determined based on the spatial angle.

[0011] As a preferred embodiment of the CNC feed rate dynamic adjustment method for adaptive wood cutting load described in this invention, wherein: determining the anisotropic cutting specific energy correction coefficient includes: Pre-set the unit specific energy for parallel cutting and cross cutting; When the spatial angle approaches a parallel state, the correction coefficient approaches the unit energy of parallel cutting; when the spatial angle approaches a perpendicular state, the correction coefficient approaches the unit energy of transverse cutting. Furthermore, the value of the correction coefficient exhibits a non-linear relationship with the sine or cosine value of the spatial angle.

[0012] As a preferred embodiment of the CNC feed rate dynamic adjustment method for adaptive wood cutting load described in this invention, the process of obtaining real-time net cutting power includes: Construct a no-load power mapping table for the spindle motor at different speeds; Real-time monitoring of the input voltage and current of the spindle motor, and calculation of the total active power; The corresponding no-load power is obtained by consulting the mapping table based on the current spindle speed. Subtracting the no-load power from the total active power yields the real-time net cutting power generated solely by the cutting action.

[0013] As a preferred embodiment of the adaptive CNC feed rate dynamic adjustment method for wood cutting load described in this invention, the calculation method of the normalized hardness factor includes: Calculate the ratio of real-time net cutting power to predicted theoretical cutting power; When the ratio is within a preset unit range, the wood material is determined to be uniform. When the ratio is greater than a preset threshold, high-hardness knots or high-density areas are determined to be detected. When the ratio is significantly less than the preset threshold, internal cavities or decayed areas of the wood are determined to be detected.

[0014] As a preferred embodiment of the CNC feed rate dynamic adjustment method for adaptive wood cutting load described in this invention, the generation rule of the geometric constraint adjustment amount includes: Set the rated maximum power or safe power threshold of the spindle motor; When the predicted theoretical cutting power exceeds the safe power threshold, a proportional coefficient less than 1 is generated as a geometric constraint adjustment amount. The proportionality coefficient is directly proportional to the safe power threshold and inversely proportional to the predicted theoretical cutting power, in order to limit overload caused by excessive cutting depth.

[0015] As a preferred embodiment of the CNC feed rate dynamic adjustment method for wood cutting load adaptation described in this invention, the generation rule of the physical adaptive adjustment amount includes: Construct a nonlinear penalty function for the normalized hardness factor; When the normalized hardness factor is within the normal range, the physical adaptive adjustment amount remains at the baseline value. When the normalized hardness factor exceeds the threshold characterizing the knot, the physical adaptive adjustment amount is reduced exponentially or in a higher-order polynomial manner through the nonlinear penalty function to force a reduction in the feed rate.

[0016] As a preferred embodiment of the CNC feed rate dynamic adjustment method for adaptive wood cutting load described in this invention, the method further includes a smoothing process step after the synthesized target feed rate command. This smoothing process step includes: The synthesized speed command is input into the S-type speed planner or filter. Based on the maximum acceleration and maximum jerk limits allowed by the machine tool servo system, the sudden changes in the speed command are smoothly processed to generate the final execution command sent to the driver.

[0017] Compared with existing technologies, the beneficial effects of this solution are: 1. By constructing a digital twin model of wood and conducting virtual cutting simulation within a preset look-ahead time window, the load caused by changes in cutting geometry (such as increased cutting depth and width) can be predicted in advance, thereby allowing for pre-adjustment of the feed rate. This fundamentally overcomes the technical shortcomings of traditional feedback control systems, which suffer from lag and inability to react promptly before sudden load changes, thus improving the safety of the processing.

[0018] 2. By comparing the predicted theoretical cutting power with the measured net cutting power, the normalized hardness factor is calculated. This effectively distinguishes whether changes in cutting load originate from predictable "geometric factors" or from unpredictable "physical material mutations" (such as encountering knots). It solves the problem of existing technologies being unable to identify the source of the load and only able to generally reduce speed, making the speed adjustment strategy more targeted and intelligent.

[0019] 3. Furthermore, this invention assigns a texture growth direction vector to each voxel when constructing the digital twin model, and introduces a correction coefficient related to the angle between the tool feed direction and the texture direction when calculating the predicted power. This quantifies the key physical characteristic of wood anisotropy, making the predicted theoretical power closer to physical reality and providing a solid foundation for the precise control of CNC systems. Attached Figure Description

[0020] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Wherein: Figure 1 This is a flowchart illustrating the overall process of a CNC feed rate dynamic adjustment method for wood cutting load adaptation according to an embodiment of the present invention. Detailed Implementation

[0021] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0022] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0023] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0024] This invention is described in detail with reference to the schematic diagrams. When detailing the embodiments of this invention, for ease of explanation, the cross-sectional views illustrating the device structure may be partially enlarged, not adhering to the usual scale. Furthermore, the schematic diagrams are merely examples and should not be construed as limiting the scope of protection of this invention. In actual fabrication, the three-dimensional spatial dimensions of length, width, and depth should be included.

[0025] Furthermore, in the description of this invention, it should be noted that the terms "upper," "lower," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. These terms are used solely for the convenience of describing the invention and for simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the invention. In addition, the terms "first," "second," or "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0026] Unless otherwise explicitly specified and limited, the terms "installation," "connection," and "joining" in this invention should be interpreted broadly. For example, they can refer to fixed connections, detachable connections, or integral connections; similarly, they can refer to mechanical connections, electrical connections, or direct connections, or indirect connections through an intermediate medium, or internal connections between two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0027] Example 1 Reference Figure 1 This is the first embodiment of the present invention, which provides a method for dynamically adjusting the CNC feed rate to adapt to wood cutting load, including: S1. Construct a digital twin model of the wood to be processed in the CNC system, and discretize the wood entity into a three-dimensional voxel matrix containing physical properties.

[0028] Furthermore, a digital twin model of the wood to be processed is constructed in the CNC system, and the wood blank is non-destructively scanned to obtain its internal three-dimensional structural data.

[0029] Preferably, Industrial Computed Tomography (CT) technology is employed. The wood to be processed is scanned from multiple angles using a CT scanner, obtaining a series of two-dimensional tomographic grayscale images (slices). It is important to emphasize that the grayscale value of each pixel in these images directly reflects the attenuation coefficient of the material as X-rays penetrate that point, and this coefficient is closely related to the material's density and atomic number. Therefore, CT grayscale values ​​can serve as a fundamental physical quantity characterizing the local density of the wood. Subsequently, using filtered back-projection or iterative reconstruction algorithms, the acquired two-dimensional slice sequence is reconstructed into a three-dimensional scalar field in a computer to reproduce the density distribution within the wood blank.

[0030] Furthermore, the reconstructed three-dimensional continuous density field is discretized, a process known as voxelization. This process divides the entire envelope space of the wood blank into a regular three-dimensional mesh, where each cubic unit is a voxel. The resolution of a voxel is determined by the side length of a single voxel. The selection of the side length of this single voxel requires a trade-off between ensuring a minimum size that can characterize typical defects such as knots and cracks inside the wood and meeting the efficiency requirements of real-time look-ahead calculations in the CNC system. An excessively large... This can lead to insufficient model accuracy and an inability to identify local hard points; while an excessively small... This would lead to a dramatic increase in data volume, posing a challenge to computing resources and processing speed. Each voxel It is assigned an average density value calculated from the average value or center point value of the CT data within its occupied space. .

[0031] Furthermore, each voxel unit in the three-dimensional voxel matrix is ​​assigned a set of structured physical property data. This process aims to expand single density information into a multi-dimensional description of mechanical properties, where the physical property data includes the following: Location status indicator Used to distinguish whether a voxel belongs to solid material or background space (such as cavities inside wood or air outside an outer contour). This is achieved by setting a density threshold. The density value of each voxel is then determined. If... If so, the voxel is determined to be an entity, and its position status is identified. If it is set to 1, it is otherwise determined to be invalid or an empty voxel. It is set to 0. By setting a density threshold, it is ensured that cutting calculations are performed only on actual existing material.

[0032] Texture growth direction unit vector Used to characterize the anisotropy of wood. Since the cutting force of wood is closely related to the cutting angle of the tool relative to the fiber direction, it is necessary to calculate the fiber orientation of the local region of each solid voxel using image processing algorithms to quantify this effect.

[0033] Specifically, an analysis method based on three-dimensional structure tensors can be used. For each voxel... In one of its neighborhood windows Internal calculation of 3D gradient And construct the structure tensor matrix. : in, It is a Gaussian weighted function. These are points within the neighborhood. For the matrix... Eigenvalue decomposition is performed, and the eigenvector corresponding to the smallest eigenvalue represents the direction in which the gradient change is slowest in that local region. Since the fiber structure of wood exhibits continuity and consistency along its growth direction, the change in CT density values ​​is most gradual in this direction. Therefore, this eigenvector can stably characterize the physical texture direction of wood, physically corresponding to the growth direction of wood fibers. After normalizing this eigenvector, the unit vector of the texture growth direction is obtained. .

[0034] It should be noted that the principle for selecting the size of the neighborhood window is that its physical size should be comparable to the side length of several voxels and be able to cover a typical cycle or characteristic size of the wood fiber structure to ensure the accuracy of local texture direction while suppressing noise. For example, if the voxel resolution is 0.5 mm, a 5×5×5 or 7×7×7 voxel cube window is usually preferred, and technicians can adapt it according to the texture fineness of the specific wood.

[0035] Preset reference hardness value : The basic material hardness used to characterize voxels. Since the hardness of wood is usually strongly correlated with its density, an empirical relationship model between density and hardness can be established through prior material mechanics experiments (e.g., Janka hardness testing on wood samples of different density grades). It is important to emphasize that this relationship model can be a lookup table or a nonlinear mapping function. Based on this, the reference hardness value of each voxel can be calculated from its density value: This benchmark value represents the theoretical hardness without considering abnormal conditions such as knots.

[0036] Furthermore, to support efficient Boolean intersection operations between the virtual tool sweep volume and the voxel matrix, an optimized data structure is needed to store this massive three-dimensional voxel matrix. Considering that a large number of voxels will be removed after the wood blank is processed (i.e., ... (The voxel matrix becomes 0), exhibiting sparsity. Therefore, the present invention preferably employs an octree structure for storage. It should be explained that an octree can hierarchically partition space, compressing voxel regions with similar properties (such as large, uniform pieces of wood or external air) into a single node, greatly reducing storage overhead. Furthermore, its tree structure enables rapid retrieval and access of voxels within a spatial region, thereby ensuring real-time computing performance.

[0037] S2. Within the preset look-ahead time window before the actual cutting occurs, a sweep volume of the virtual tool is generated according to the tool path instructions of the CNC machining program, and the Boolean intersection operation between the sweep volume and the three-dimensional voxel matrix is ​​performed to extract the set of voxels to be removed.

[0038] It should be noted that by virtually interacting the CNC system's digital instructions (G-code) with the workpiece's physical property model (digital twin voxel matrix), the upcoming cutting behavior can be accurately predicted before the tool actually contacts the workpiece, fundamentally overcoming the lag problem of traditional feedback control.

[0039] Furthermore, the CNC system's interpolator, when executing the current G-code program segment, pre-reads and parses several subsequent program segments. This invention utilizes this mechanism to set a preset look-ahead time window. .

[0040] Furthermore, this forward-looking time window The setting is a trade-off. If If the time is too short, the system's reaction time will be insufficient, making smooth speed regulation impossible and degenerating into quasi-reactive control; if... Excessive length of the prediction path increases the computational burden of each prediction, affecting the system's real-time performance. Furthermore, an excessively long prediction path may deviate from the actual future situation due to accumulated errors during actual processing. Therefore, It is typically set to the sum of several to dozens of interpolation cycles of the CNC system. In this embodiment, its typical value is between 50ms and 200ms, ensuring that it can effectively anticipate load changes (such as entering a large cutting depth area or encountering knots) and correct predictions in a timely manner during dynamic changes.

[0041] It should be noted that within this time window, the system extracts the path instruction sequence that the tool center point (TCP) will travel from the G-code buffer, including linear interpolation (G01) and circular interpolation (G02 / G03), etc.

[0042] Furthermore, the geometry of the cutting tool is combined with its motion trajectory within a time window to generate a continuous spatial envelope, also known as the virtual tool's swept volume, denoted as... .

[0043] Specifically, the input retrieves the 3D model or key geometric parameters of the currently used tool from the CNC system's tool library. For example, for a flat end mill, this is the radius and height; for a ball end mill, it is the ball end radius. Input time window. TCP path sequence within ,in, .

[0044] Furthermore, since the generation of the swept volume is a computational geometry problem, the space swept by the tool geometry along each tiny segment of the path is calculated and accumulated.

[0045] Furthermore, for a linear interpolation segment, the swept volume is a generalized cylinder. For example, a cylinder with a radius of... The flat-bottom end mill moves along a straight segment, and its swept volume is a cylinder with flat ends. For the circular interpolation segment, the swept volume is part of a torus. By performing a Boolean union operation on the swept volumes of all micro-segments along the look-ahead path, a complete sweep volume representing the future can be formed. Within a given time, the cutting tool will occupy and remove the three-dimensional space of the material. .

[0046] Furthermore, perform virtual tool scanning. With the constructed three-dimensional voxel matrix The Boolean intersection operation between them. Its goal is to find all voxels that simultaneously satisfy the following two conditions: the first condition is that their spatial location is located at... Internal; the second condition is: its location status identifier. It is 1 (i.e., solid material).

[0047] Furthermore, to meet real-time requirements, this operation does not traverse all voxels, but instead utilizes a spatial index structure (such as an octree) to greatly accelerate the retrieval process. The process is as follows: Calculate the sweep volume The axis-aligned bounding box (AABB) is used. Utilizing an octree structure, all tree nodes intersecting with the AABB can be retrieved extremely quickly. This step rapidly eliminates a large number of voxels that have no spatial relation to the cutting region, reducing the computational scope by several orders of magnitude. For the roughly selected candidate voxels, a precise geometric interior / exterior point determination is then performed. That is, it is determined whether the center point or eight vertices of each candidate voxel are located within the sweep volume. Inside the geometric model. For voxels that pass the judgment, check their positional status identifiers. Only when Only then is the voxel identified as the material to be cut. All voxels that meet the conditions... Together they constitute the set of voxels to be removed, denoted as .

[0048] It should be noted that the obtained set of voxels to be removed This set not only contains the geometric information of the material to be cut, but each voxel in the set also carries its own physical property data (such as texture direction vectors). and reference hardness value This information serves as direct input for calculating the predicted theoretical cutting power. Simultaneously, the instantaneous material removal rate (MRR) is calculated by statistically analyzing the number of voxels in this set to directly reflect the geometric severity of the cutting. This removal rate is expressed as: in, It is a set The number of middle voxels, It is the side length of a single voxel.

[0049] S3. Based on the sum of the geometric volumes of the specific voxel sets to be removed, and combined with the effect of texture direction on the correction of cutting resistance, calculate the predicted theoretical cutting power within the time window.

[0050] It should be noted that cutting power is physically equal to the work done in removing material per unit time, and its classical model can be expressed as the product of specific cutting energy and material removal rate (MRR). However, for heterogeneous materials such as wood, specific cutting energy is not a constant. The present invention analyzes the properties of each voxel to be removed, calculates a customized effective specific cutting energy for each voxel, and then accumulates these values ​​to obtain a high-precision total power prediction.

[0051] Furthermore, in order to quantify the influence of wood grain orientation on cutting resistance, the set of voxels to be removed was... Each voxel in The following operations need to be performed independently: S301. Obtain the composite motion velocity vector of the tool center point within the current look-ahead time window from the interpolator of the CNC system. This vector represents the main motion direction of the tool cutting edge. Simultaneously, obtain the velocity vector from the voxel. In the structured attribute data, the texture growth direction unit vector assigned in step S1 is read directly.

[0052] S302. Calculate the spatial angle between the two vectors mentioned above. This can be achieved through the vector dot product operation: in, (Usually, an acute angle is used for analysis). This included angle Intuitively quantifying voxels The location and the relationship between the cutting direction of the tool and the direction of the wood fibers. This indicates a tendency towards parallel cutting, while This indicates a cutting pattern approaching the horizontal grain.

[0053] S303, Establish a spatial angle A nonlinear function mapped to the correction factor value is used to determine the anisotropic cutting specific energy correction factor. Specifically, through preliminary cutting experiments, the typical unit specific energy for parallel cutting was calibrated. Energy density per unit of cross-cutting These two values ​​are fundamental physical constants describing the mechanical properties of specific wood species under specific processing conditions (such as tool geometry, rotation speed, etc.), and are generally... Then, an interpolation function is constructed such that when the included angle... When it changes, the correction factor can and A smooth transition occurs between them. In this embodiment, a preferred function model that conforms to the limitation of "non-linear variation with the sine or cosine value of the spatial angle" can be expressed as: in, It is an exponent greater than 1 (e.g., This is used to adjust the nonlinearity of the transition curve, making it more closely match the actual variation of cutting force.

[0054] It should be noted that when (Pure parallel cutting) ,but .when (Pure cross-cutting) ,but Therefore, for any intermediate angle, the function model can provide a nonlinear weighted average, thus accurately simulating the physical process from an easily machinable parallel-grooving state to a cross-grooving state with significantly increased resistance.

[0055] Furthermore, after calculating the anisotropic cutting energy correction coefficient for each voxel to be removed, the contributions of all voxels are summed to obtain the total predicted theoretical cutting power over the entire look-ahead time window.

[0056] Specifically, for sets any voxel in The theoretical power required to remove it (i.e., the power contribution of a single element) can be expressed as: in, It was just calculated and is a voxel. Customized unit cutting energy. It is a voxel The preset baseline hardness value is used as a dimensionless multiplicative factor to correct the basic unit specific energy, so as to reflect the difference in basic hardness caused by density changes. It is the volume of a single voxel (e.g., ), representing the volume of material removed from that voxel. It is the length of the forward-looking time window. This represents the contribution of the voxel to the average material removal rate.

[0057] Specifically, the set of voxels to be removed The final predicted theoretical cutting power can be obtained by summing the power contributions of all voxels. (i.e., total power): S4. During the actual machining process, the real-time drive signal of the spindle motor is collected synchronously, and the net cutting power is obtained after filtering out the no-load loss.

[0058] Furthermore, even when the spindle motor is not in contact with the workpiece and is rotating under no-load, it still consumes power. This power is mainly used to overcome bearing friction, air resistance, and internal copper and iron losses within the motor. This no-load power... It is not a constant value; it varies with the spindle speed. This exhibits a strong positive correlation (usually a quadratic or higher power relationship). If this speed-related loss is not eliminated, the motor's own energy consumption will be misjudged as the cutting load, leading to serious deviations in subsequent calculations. Therefore, before any actual cutting operations, the spindle motor's energy consumption characteristics need to be calibrated offline once to construct an idle power model.

[0059] Specifically, during the experiment, no tool was mounted on the spindle, or only a tool was mounted without contacting any workpiece. Then, via CNC system commands, the spindle was controlled to operate stably at a series of discrete speed points (e.g., every 500 RPM from the lowest operating speed to the highest rated speed). At each stable speed point... Below, the stable active power consumed by the motor is measured and recorded using the monitoring function built into a high-precision power analyzer or the motor driver, and denoted as . Finally, all the collected data were compared with... Stored in this way, forming a lookup table (LUT), denoted as This mapping table represents the no-load power mapping of the spindle motor at different speeds. Furthermore, to improve query accuracy, a polynomial fitting of these data points can be used to obtain a continuous function. .

[0060] Furthermore, in actual cutting processes, it is necessary to monitor the electrical parameters of the drive spindle motor frequently.

[0061] Specifically, the instantaneous voltage of the three-phase AC power is synchronously acquired by high-speed voltage and current sensors installed at the input of the spindle driver. and current These signals are sent to the real-time computing unit of the CNC system via a data acquisition card (DAQ) at a sampling rate in the kHz range. The acquired voltage and current signals are processed to calculate the total active power of the spindle motor. For a three-phase balanced system, the calculation formula is: in, This is the effective value (RMS) of the line voltage. This represents the effective value (RMS) of the line current. The power factor is represented by the cosine of the phase difference between voltage and current. It is important to emphasize that these values ​​can be calculated in real time by performing digital signal processing (such as Fast Fourier Transform, FFT) on instantaneous signals within one or more power grid cycles.

[0062] Furthermore, by combining the real-time measured total power with the pre-calibrated no-load power model, the net cutting power is finally separated.

[0063] Specifically, after calculating the total active power at the current moment... Simultaneously, the commanded spindle speed or the actual speed fed back by the encoder is read from the CNC system. .by As an index, in the constructed no-load power mapping table Find the corresponding no-load power value in the [database / system]. If If the value falls exactly between the two calibration points, then linear interpolation or a higher-order interpolation method is used to calculate the accurate no-load power. This ensures smoothness and accuracy. Then, the total active power is subtracted from the no-load power corresponding to the current rotational speed; the difference is the power consumed entirely by the interaction between the tool and the workpiece for material removal, i.e., the real-time net cutting power. : Furthermore, due to minute vibrations during electrical measurement and cutting processes, the calculated... The signal may contain high-frequency noise. To obtain a stable and reliable feedback value, it is usually necessary to perform digital low-pass filtering, such as using a moving average filter or a first-order Butterworth filter, to smooth the signal and remove glitches. This yields a real-time net cutting power that dynamically and accurately reflects the actual load during the cutting process.

[0064] S5. Compare the real-time net cutting power with the predicted theoretical cutting power to calculate the normalized hardness factor that reflects the properties of wood material, thereby decoupling the geometric cutting component and the physical material component in the load.

[0065] It should be noted that the calculated predicted theoretical cutting power The theoretical load is derived from known geometric information (number of voxels to be removed) and prior physical properties (baseline hardness, texture orientation) in the digital twin model. Therefore, This can be considered a quantitative representation of the geometric cutting components. Specifically, it represents the load magnitude determined solely by the toolpath and cutting parameters under ideal conditions where the material is perfectly homogeneous and conforms to the model's preset parameters. The obtained real-time net cutting power... This is the total cutting load that actually occurs in the physical world, encompassing all influencing factors (including unknown material anomalies). If the actual hardness of the material perfectly matches the preset reference hardness in the digital twin model, then... Should be with The numerical results show a high degree of agreement; any deviation must stem from unforeseen changes in the physical material components. Therefore, to quantify the degree of this deviation, this invention defines a dimensionless diagnostic index, namely the normalized hardness factor, denoted as... It is calculated from the ratio of real-time net cutting power to predicted theoretical cutting power: It is important to note that in practical applications, to avoid air cutting or extremely small cutting amounts (which would lead to...) A calculation error occurred when the denominator was zero; a small positive threshold needs to be set. .when At that time, the system defaults to normal material conditions and directly orders... .

[0066] Furthermore, through the study of The real-time value is used to determine the range and classify the wood material properties of the current cutting area in real time. In this embodiment, it is demonstrated through the following three cases: Scenario 1: The wood has a uniform texture ( ) Judgment rule: When When the value is within a preset unit range, for example, (in, If there is a small tolerance, such as 0.15, used to filter out normal signal fluctuations, then it is determined that the wood material in the current cutting area is uniform and its actual hardness matches the preset value in the digital twin model.

[0067] It should be noted that when situation one occurs, it also means... The actual load matches the theoretical prediction. At this point, the fluctuation of the cutting load is mainly dominated by geometric factors, and the prediction of the digital twin model is accurate.

[0068] Scenario 2: An abnormally high hardness was detected ( ) Judgment rule: When The value is greater than a preset high hardness threshold. At times (for example, If the cutting tool is currently cutting a high-hardness nodule or a high-density area, then it is determined that the tool is currently cutting a high-hardness nodule or a high-density area.

[0069] It should be noted that when encountering situation two, it means... Significantly greater than In cutting geometry (by...) Given the known situation, the only explanation is that the material's unit cutting energy far exceeded expectations, indicating the encounter of hard particles. The value directly reflects the hardness multiple of that region relative to normal wood.

[0070] Scenario 3: Low hardness or defects detected ( ) Judgment rule: When The value is less than a preset low hardness threshold. At times (for example, If the tool encounters a cavity, crack, or decayed area inside the wood, it is determined that the tool has encountered such an area.

[0071] It should be noted that when encountering situation two, the explanation... much smaller In other words, it was predicted that there should be material to be cut and corresponding power generated at this location, but in reality, not that much energy was consumed, indicating that the area is missing material or the material is extremely porous.

[0072] S6. Based on the geometric constraint adjustment amount generated by the predicted theoretical cutting power and the physical adaptive adjustment amount generated by the normalized hardness factor, the target feed speed command is synthesized together to dynamically control the CNC machine tool.

[0073] It should be noted that the main objective of the geometric constraint adjustment is to achieve proactive overload protection, based on the calculated predicted theoretical cutting power. This allows the CNC system to anticipate potential power spikes before the tool enters the high-cutting-depth region. By actively reducing the feed rate, it ensures that the actual machining power remains within the safe operating range of the spindle motor, thereby protecting the motor and tool and preventing chipping or stalling caused by a sharp increase in cutting depth or width. The steps for generating this geometric constraint adjustment are as follows: Set the safe power threshold for the spindle motor This threshold is typically 75% to 85% of the spindle's rated power to allow for sufficient safety margin in case of unforeseen circumstances. For example, for a spindle with a rated power of 10kW, it can be set to... Based on this, the geometric constraint adjustment amount It can be designed as a range in The proportionality coefficient between them is calculated using the following logic: in, The input is the predicted theoretical cutting power. Preset safe power threshold. A regulation index greater than 0 (usually taken as...) ), used to adjust the sensitivity of the adjustment.

[0074] It should be noted that when hour, This means that, from a geometric point of view, even at the original velocity set in the G-code... Processing will not result in overload, therefore no restrictions are imposed. hour, It is a proportionality coefficient less than 1. This proportionality coefficient is related to the safe power threshold. Proportional ( The larger the coefficient, the higher the predicted theoretical cutting power. Inversely proportional ( The larger the coefficient, the smaller the coefficient. This inverse relationship ensures that the more severe the predicted overload, the stronger the applied speed suppression, thus effectively clamping the potential future power near the safety line. Meanwhile, the exponent... The introduction of this allows the inhibition relationship to be non-linear, for example, when At that time, the initial inhibitory effect is stronger and the reaction is more rapid.

[0075] It should be noted that, unlike the geometric constraint adjustment amounts mentioned above, the goal of the physically adaptive adjustment amounts is to achieve a real-time response to abrupt changes in material properties. This is achieved by utilizing the obtained normalized hardness factor. As input, it is specifically designed to address material anomalies not anticipated by the digital twin model, especially extremely hard knots. When hard points are detected, decisive and drastic speed reduction measures must be taken to prevent impact damage to the tool and scorching marks on the workpiece surface. The steps for generating this physically adaptive adjustment are as follows: Set a hardness threshold to characterize the knot. This value is greater than the upper limit used in step S5 to determine material uniformity (i.e., For example, it can be set. With step S5 Maintain consistency. This includes the physical adaptive adjustment amount. Similarly, within a range The proportionality coefficient between them is calculated using the following logic: in, This is the input normalized hardness factor. This is the preset threshold for knot hardness. It is a positive penalty coefficient used to control the severity of the speed decrease. The larger the size, the faster the deceleration.

[0076] It should be noted that when hour, This indicates that the material hardness is within a normal or acceptable range, and there is no need to adjust the speed due to material properties, thus allowing the physical adaptive adjustment amount to remain at the baseline value. When When, the penalty function is activated. When the threshold is just exceeded, the rate of decrease is relatively mild; but as the threshold is exceeded, the rate of decrease slows down. The further increase, The term will cause the exponential part to increase rapidly. It will decrease exponentially, rapidly approaching 0, causing the physical adaptive adjustment amount to decrease exponentially or in a higher-order polynomial manner. This enables an extremely sensitive and powerful "braking" response to dangerous hard points, forcibly reducing the feed rate and ensuring machining safety.

[0077] Furthermore, the raw feed rate programmed in the G code... By multiplying and adding the two adjustment values ​​mentioned above, the initial target feed rate can be obtained. : It is important to emphasize that using a multiplication structure ensures that the constraints of any channel (geometric or physical) directly affect the final velocity, thus conforming to the "AND" relationship of the control logic.

[0078] In addition, directly output It may be because or Rapid changes in speed can create abrupt or sharp inflection points, exceeding the dynamic response capabilities of the machine tool servo system and causing vibration and shock. Therefore, smoothing is necessary, as follows: S501, Combine the speed command sequence The input is fed into an S-type velocity planner or a digital low-pass filter with similar functionality.

[0079] S502, Set the maximum allowable acceleration of the machine tool servo system for this planner. and maximum jerk These parameters are typically provided by the machine tool manufacturer and are key indicators characterizing the dynamic performance of the machine tool.

[0080] The S503 and S-type speed planners will automatically calculate a path that smoothly transitions from the current speed to the target speed while satisfying the above constraints. The velocity curve. It is important to emphasize that the characteristic of this velocity curve is that its acceleration curve is trapezoidal, and the jerk is a piecewise constant, thus ensuring the continuity of velocity and acceleration and avoiding impact.

[0081] S504, the smoothed speed command output by the planner. This refers to the execution instructions ultimately sent to the servo driver.

[0082] It should be noted that, through the above processing, the present invention can ultimately generate a dynamic feed rate curve that maximizes processing efficiency and ensures safe, stable and high-quality processing, thus perfectly realizing the adaptive load control of wood cutting.

[0083] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of this application can be implemented using various computer languages, such as the object-oriented programming language Java and the interpreted scripting language JavaScript.

[0084] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0085] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0086] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0087] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0088] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A method for dynamically adjusting CNC feed rate to adapt to wood cutting load, characterized in that, include: A digital twin model of the wood to be processed is constructed in the CNC system, and the wood entity is discretized into a three-dimensional voxel matrix containing physical properties. Within a preset look-ahead time window before the actual cutting occurs, a sweep volume of a virtual tool is generated according to the tool path instructions of the CNC machining program, and the Boolean intersection operation between the sweep volume and the three-dimensional voxel matrix is ​​performed to extract the set of voxels to be removed. Based on the total geometric volume of the set of voxels to be removed, and considering the effect of texture direction on the correction of cutting resistance, the predicted theoretical cutting power within the time window is calculated. During actual machining, the real-time drive signal of the spindle motor is collected synchronously, and the real-time net cutting power is obtained after filtering out the no-load loss. The real-time net cutting power is compared with the predicted theoretical cutting power to calculate the normalized hardness factor that reflects the properties of wood material, thereby decoupling the geometric cutting component and the physical material component in the load. The geometric constraint adjustment amount generated by the predicted theoretical cutting power and the physical adaptive adjustment amount generated by the normalized hardness factor are used to synthesize the target feed rate command and dynamically control the CNC machine tool.

2. The CNC feed rate dynamic adjustment method for adaptive wood cutting load as described in claim 1, characterized in that, Each voxel unit in the three-dimensional voxel matrix is ​​assigned structured attribute data, which includes: The location status identifier representing the presence or absence of the voxel, the unit vector representing the direction of wood grain growth at the location of the voxel, and the preset reference value representing the hardness of the base material of the voxel. The three-dimensional voxel matrix is ​​stored using a spatial index structure to support fast retrieval.

3. The CNC feed rate dynamic adjustment method for adaptive wood cutting load as described in claim 2, characterized in that, The Boolean intersection operation includes: A continuous spatial envelope is generated based on the tool motion trajectory and tool geometry within the current interpolation cycle. A spatial index structure is used to retrieve voxels located within the spatial envelope and whose position status is identified as existing. These voxels are marked as the cutting objects at the current moment, and their total volume is calculated to obtain the instantaneous material removal rate.

4. The CNC feed rate dynamic adjustment method for adaptive wood cutting load as described in claim 2, characterized in that, The calculation method for the effect of the texture direction on the correction of cutting resistance includes: For each voxel in the set of voxels to be removed, the current tool feed synthesis speed vector and the texture growth direction unit vector recorded by the voxel are obtained respectively. Calculate the spatial angle between the feed synthesis velocity vector and the texture growth direction unit vector; The anisotropic cutting energy correction coefficient is determined based on the spatial angle.

5. The CNC feed rate dynamic adjustment method for adaptive wood cutting load as described in claim 4, characterized in that, The determination of the anisotropic cutting specific energy correction coefficient includes: Pre-set the unit specific energy for parallel cutting and cross cutting; When the spatial angle approaches a parallel state, the correction coefficient approaches the unit energy of parallel cutting; when the spatial angle approaches a perpendicular state, the correction coefficient approaches the unit energy of transverse cutting. Furthermore, the value of the correction coefficient exhibits a non-linear relationship with the sine or cosine value of the spatial angle.

6. The CNC feed rate dynamic adjustment method for adaptive wood cutting load as described in claim 1, characterized in that, The process of obtaining real-time net cutting power includes: Construct a no-load power mapping table for the spindle motor at different speeds; Real-time monitoring of the input voltage and current of the spindle motor, and calculation of the total active power; The corresponding no-load power is obtained by consulting the mapping table based on the current spindle speed. Subtracting the no-load power from the total active power yields the real-time net cutting power generated solely by the cutting action.

7. The CNC feed rate dynamic adjustment method for adaptive wood cutting load as described in claim 1, characterized in that, The calculation method for the normalized hardness factor includes: Calculate the ratio of real-time net cutting power to predicted theoretical cutting power; When the ratio is within a preset unit range, the wood material is determined to be uniform. When the ratio is greater than a preset threshold, high-hardness knots or high-density areas are determined to be detected. When the ratio is significantly less than the preset threshold, internal cavities or decayed areas of the wood are determined to be detected.

8. The CNC feed rate dynamic adjustment method for adaptive wood cutting load as described in claim 1, characterized in that, The generation rules for the geometric constraint adjustment amount include: Set the rated maximum power or safe power threshold of the spindle motor; When the predicted theoretical cutting power exceeds the safe power threshold, a proportional coefficient less than 1 is generated as a geometric constraint adjustment amount. The proportionality coefficient is directly proportional to the safe power threshold and inversely proportional to the predicted theoretical cutting power, in order to limit overload caused by excessive cutting depth.

9. The CNC feed rate dynamic adjustment method for wood cutting load adaptation as described in claim 7, characterized in that, The rules for generating the physical adaptive adjustment amount include: Construct a nonlinear penalty function for the normalized hardness factor; When the normalized hardness factor is within the normal range, the physical adaptive adjustment amount remains at the baseline value. When the normalized hardness factor exceeds the threshold characterizing the knot, the physical adaptive adjustment amount is reduced exponentially or in a higher-order polynomial manner through the nonlinear penalty function to force a reduction in the feed rate.

10. The CNC feed rate dynamic adjustment method for adaptive wood cutting load as described in claim 1, characterized in that, Following the synthetic target feed rate command, a smoothing process step is also included, which includes: The synthesized speed command is input into the S-type speed planner or filter. Based on the maximum acceleration and maximum jerk limits allowed by the machine tool servo system, the sudden changes in the speed command are smoothly processed to generate the final execution command sent to the driver.

Images