A multi-tool head collaborative cutting method and device of distributed closed-loop control

Abstract

The application provides a multi-head cooperative cutting method and device of distributed closed-loop control. The cutting path atlas and initial compensation parameter set generated by a master control node are distributed to sub-controller nodes arranged at multiple cutting heads in real time. A deviation vector between multi-dimensional sensing features at the current time and preset path features is calculated. The Z-axis movement position and X / Y-axis movement speed of the corresponding head are locally compensated and adjusted according to the deviation vector. The deviation vector change caused by mechanical disturbance during the local compensation and adjustment is calculated. The deviation vector change is synchronously transmitted to adjacent sub-controller nodes in real time, so that the adjacent sub-controller nodes implement movement path correction according to the deviation vector change. Through the distributed differential information feedback mechanism, the physical coupling interference between the multiple heads is converted into preventive global synchronous compensation action, thereby realizing high consistency of the cutting seam geometric shape under complex working conditions.

Description

Technical Field

[0001] This invention relates to the field of automation control technology, specifically to a multi-blade collaborative cutting method and device based on distributed closed-loop control. Background Technology

[0002] In modern high-precision cutting of metals or composite materials such as thin films and composite sandwich materials, the integrated solution of multi-head parallel operation has become the mainstream trend due to the increasing demand for production efficiency. However, in actual processing, the physical properties of the material itself, such as elasticity and viscoelasticity, as well as the disturbance of cutting force and the accumulation of thermal effects during processing, cause complex dynamic deformation of the material along the cutting path. Traditional closed-loop control systems drive a single cutter to perform local correction actions when kerf deviation is detected. However, this compensation behavior itself becomes a new and dynamic source of mechanical disturbance. Therefore, multi-head cutting can cause unpredictable, large-scale and random dynamic deformation of the material in space. This untrackable chain deformation can lead to uncontrolled kerf width, edge tearing, and even material scrap, seriously affecting the processing accuracy of multi-head cutting in the manufacturing field. Summary of the Invention

[0003] Based on the above problems, this invention proposes a distributed closed-loop control method and device for multi-head collaborative cutting. Through a distributed differential information feedback mechanism, the physical coupling interference between multiple cutters is transformed into a preventive global synchronous compensation action, thereby achieving a high degree of consistency in the geometry of the cut under complex working conditions.

[0004] In view of this, the first aspect of the present invention proposes a distributed closed-loop control multi-blade collaborative cutting method, comprising: Generate the cutting path map of the cutting object and the initial compensation parameter set for each cutter head region; The cutting path map and initial compensation parameter set are distributed in real time to the sub-controller nodes deployed on multiple cutting heads; Real-time acquisition of multi-dimensional sensing features during the cutting process; Calculate the deviation vector between the multi-dimensional perception features at the current moment and the preset path features; The Z-axis position and X / Y-axis speed of the corresponding tool head are locally compensated and adjusted according to the deviation vector. Calculate the change in the deviation vector caused by mechanical disturbance during the execution of the local compensation adjustment; The deviation vector change is synchronized to the adjacent sub-controller nodes in real time, so that the adjacent sub-controller nodes can perform motion path correction based on the deviation vector change.

[0005] Optionally, the steps for generating the initial cutting path map and the initial compensation parameter set for each cutter head region specifically include: Construct a digital twin model of the object to be cut; Based on the aforementioned digital twin model, the stress field changes generated during the cutting process are simulated in virtual space; Pre-generate dynamic stress field maps for future path points; The spatial region of the digital twin model of the object to be cut is decomposed into several sub-regions by slicing. The initial compensation parameter set is obtained by mapping the regional features of each sub-region to the parameter compensation space.

[0006] Optionally, the step of mapping the regional features of each sub-region to the parameter compensation space to obtain the initial compensation parameter set specifically includes: Calculate the first Z-axis height compensation parameters for each sub-region: , in For vector fields of Axial components, It is a function of average value; Calculate the first X / Y axis motion velocity gain for each sub-region: , in For the first The predicted strain of the material of the cut object along the tensile direction within each sub-region; Calculate the first Local pressure regulation coefficient for each sub-region: , in For the first The predicted maximum normal stress in each sub-region For pre-configured control law functions; Using the Z-axis height compensation parameters The X / Y axis motion velocity gain and the local pressure regulation coefficient Construct the initial compensation parameter set: .

[0007] Optionally, the control law function For an exponential buffer function, calculate the th Local pressure regulation coefficient of each sub-region The specific steps are as follows: , in This is the reference pressure gain coefficient under standard operating conditions. The pre-configured exponential decay coefficient.

[0008] Optionally, the step of calculating the deviation vector between the multi-dimensional perceptual features at the current moment and the preset path features specifically includes: Based on the transient stress wave signal generated by microcracks inside the material Calculate the first deviation vector: , in These are the weighting coefficients; Based on the actual pressure center coordinates during the cutting process Calculate the second deviation vector: , in These are the predicted pressure center coordinates pre-configured in the cutting path map; Based on the actual kerf depth change rate during the cutting process Calculate the third deviation vector: , in It is the predicted cut depth variation rate pre-configured in the cutting path map.

[0009] Optionally, the weighting coefficients It is a Gaussian distribution function centered at the peak of the pulse within a set time window: , Among them, in the above Gaussian distribution function is the standard deviation of the Gaussian distribution, and is a hyperparameter used to adjust the width of the feature extraction window.

[0010] Optionally, the step of locally compensating and adjusting the Z-axis movement position and X / Y-axis movement speed of the corresponding tool head based on the deviation vector specifically includes: Calculate the attenuation coefficient of the X / Y axis motion velocity: , in, The first deviation vector The length of the mold, This is a preset damage sensitivity attenuation constant; Calculate the incremental compensation value for the Z-axis motion position: , in, The second deviation vector The length of the mold, The second deviation vector The angle with the positive Z-axis; The Z-axis incremental compensation amount is determined based on the decision mapping function. The plus or minus sign; The incremental compensation value based on the Z-axis motion position The attenuation coefficient of the X / Y axis motion speed, and the controller drives the actuator to adjust the parameters of the tool head position and motion speed.

[0011] Optionally, the step of calculating the change in the deviation vector caused by the mechanical disturbance during the execution of the local compensation adjustment specifically includes: Construct the deviation vector sequence between the current time and the previous sampling time: .

[0012] Calculate the time-domain difference change of the deviation vector sequence: .

[0013] Calculate the scalar value of the deviation disturbance intensity caused by mechanical disturbance: .

[0014] Optionally, the step of synchronizing the deviation vector change to adjacent sub-controller nodes in real time, so that the adjacent sub-controller nodes can perform motion path correction based on the deviation vector change, specifically includes: Perform differential information broadcasting based on the EtherCAT protocol so that adjacent sub-controller nodes can adjust according to the time-domain differential change. and / or the deviation disturbance intensity scalar Perform local forecast data updates; Neighboring sub-controller nodes recalculate their own compensation parameter sets based on the updated local prediction data.

[0015] A second aspect of the present invention provides a distributed closed-loop controlled multi-head collaborative cutting device, comprising: The master node is used to execute global task decomposition and pre-compensation planning, and to simulate and generate the initial cutting path map and the initial compensation parameter set for each cutter head region based on the material physical properties of the cutting object. The distributed sub-controller node array is deployed on multiple interconnected cutting heads. Each sub-controller node corresponds to one cutting head. Each sub-controller node integrates a localized deformation prediction model, which is used to receive initial parameters issued by the master node and perform local closed-loop compensation based on real-time sensing data. A multi-dimensional sensor array, integrated on each of the sub-controller nodes, is used to collect sound, light, and pressure information of the object being cut during the cutting process; A high-bandwidth communication bus is connected to the master control node and the sub-controller nodes to enable real-time command issuance and broadcast synchronization of deviation vector information between sub-controller nodes. The multi-blade collaborative cutting device is configured to implement the distributed closed-loop control multi-blade collaborative cutting method according to any one of the first aspects of the present invention.

[0016] This invention proposes a distributed closed-loop control method and device for multi-cutter head collaborative cutting. The method involves distributing the cutting path map and initial compensation parameter set generated by the master control node to sub-controller nodes deployed on multiple cutting cutters in real time. It calculates the deviation vector between the multi-dimensional sensing features and the preset path features at the current moment, and performs local compensation adjustments on the Z-axis position and X / Y-axis speed of the corresponding cutter head based on the deviation vector. It also calculates the deviation vector change caused by mechanical disturbances during the local compensation adjustment and synchronizes this change to adjacent sub-controller nodes in real time. This allows adjacent sub-controller nodes to correct their motion paths based on the deviation vector change. In other words, through a distributed differential information feedback mechanism, the physical coupling interference between multiple cutters is transformed into a preventative global synchronous compensation action, thereby achieving a high degree of consistency in the cutting geometry under complex working conditions. Attached Figure Description

[0017] Figure 1 This is a flowchart of a distributed closed-loop control multi-blade collaborative cutting method provided in one embodiment of the present invention; Figure 2 This is a schematic diagram of a distributed closed-loop controlled multi-blade collaborative cutting device provided in one embodiment of the present invention. Detailed Implementation

[0018] To better understand the above-mentioned objectives, features, and advantages of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in these embodiments can be combined with each other.

[0019] 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 therefore the scope of protection of the invention is not limited to the specific embodiments disclosed below.

[0020] In the description of this invention, the term "multiple" refers to two or more. Unless otherwise explicitly defined, the terms "upper," "lower," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, and are only for the convenience of describing the invention and 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, and therefore should not be construed as a limitation of the invention. The terms "connect," "install," "fix," etc., should be interpreted broadly. For example, "connect" can be a fixed connection, a detachable connection, or an integral connection; it can be a direct connection or an indirect connection through an intermediate medium. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances. Furthermore, the terms "first," "second," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined with "first," "second," etc., may explicitly or implicitly include one or more of that feature. In the description of this invention, unless otherwise stated, "multiple" means two or more.

[0021] In the description of this specification, the terms "one embodiment," "some implementations," "specific embodiment," etc., refer to a specific feature, structure, material, or characteristic described in connection with that embodiment or example, which is included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0022] The following description, with reference to the accompanying drawings, illustrates a distributed closed-loop control multi-blade collaborative cutting method and apparatus according to some embodiments of the present invention.

[0023] like Figure 1 As shown, the first aspect of the present invention proposes a distributed closed-loop control multi-blade collaborative cutting method, comprising: Step 100: Generate the cutting path map of the cutting object and the initial compensation parameter set for each cutter head region.

[0024] In the technical solution of this invention, a finite element analysis model is used to simulate the physical properties and processing environment of the material to be cut, predict the dynamic deformation trend of the material under standard tension, and generate an initial cutting path map and an initial compensation parameter set for each cutter head region. Specifically, the system first establishes a mathematical model including the material's elastic modulus, viscoelastic parameters, and coefficient of thermal expansion. By simulating the boundary conditions under standard tension, the strain distribution map of the material on the cutting path is calculated. For example, when processing highly elastic carbon fiber fabric, the finite element analysis model predicts that when the cutter head passes through a certain area, the tension will cause a planar displacement trend of 0.5 mm in that area. Based on this prediction, the system generates an initial cutting path map containing offset compensation in advance and calculates the corresponding initial compensation parameter set for each cutter head. The above steps reduce the search space and computational difficulty of subsequent real-time control by transforming the complex physical deformation problem into preset parameters of the geometric path.

[0025] Step 200: Distribute the cutting path map and initial compensation parameter set to the sub-controller nodes deployed on multiple cutting heads in real time.

[0026] Specifically, the system employs a high-bandwidth communication bus such as EtherCAT (Ethernet Control Automation Technology) or high-performance industrial Ethernet as the data communication carrier. In the technical solution of this invention, the master control node generates and distributes cutting path maps and compensation parameter sets to each sub-controller node. These maps and parameters provide prior knowledge and benchmark references for the cutting process, transforming the complex physical deformation problem into preset parameters for the geometric path, thus reducing the search space and computational difficulty for subsequent real-time control. The actual real-time trajectory calculation is handled by each sub-controller node, avoiding the communication delay caused by large-scale sensor data transmission back to the master control node and subsequent command return.

[0027] Step 300: Real-time acquisition of multi-dimensional perceptual features during the cutting process.

[0028] Furthermore, the multi-dimensional sensing features include actual displacement features such as kerf width, kerf depth, and material surface height. The system acquires these actual displacement features during the cutting process by using laser displacement sensors and vision sensors integrated on the cutter heads of each sub-controller node. In some implementations, the laser displacement sensor is used to measure the vertical height fluctuation of the target being cut, and the vision sensor is used to capture changes in the lateral width of the kerf.

[0029] Furthermore, the multi-dimensional sensing features also include the acoustic response features of the object being cut during the cutting process. By setting an acoustic emission sensor on the cutter head of the sub-controller node, the high-frequency stress pulse signal released when the microstructure of the material inside the object undergoes abrupt changes during cutting is captured, and the acoustic response features of the object being cut during the cutting process are extracted from the high-frequency stress pulse signal.

[0030] Furthermore, the multi-dimensional sensing features also include real-time pressure features of the object being cut during the cutting process. An array of flexible tactile sensors, arranged on a cutting platform supporting the object, captures real-time pressure data exerted by the object on the flexible tactile sensors on the cutting platform as it is being cut, and extracts the pressure features of the object during the cutting process from this real-time pressure data.

[0031] Step 400: Calculate the deviation vector between the multi-dimensional perception features at the current moment and the preset path features.

[0032] Specifically, the deviation vector is the deviation between the actual feature values ​​of the object being cut and the preset feature values ​​in multiple sensory dimensions such as sound, light, and pressure during the cutting process.

[0033] Step 500: Based on the deviation vector, perform local compensation adjustment on the Z-axis movement position and X / Y-axis movement speed of the corresponding tool head.

[0034] For example, if the vision sensor detects that the kerf width has increased by 0.1 mm due to material stretching, the sub-controller immediately drives the actuator through an adaptive PID algorithm to complete the downward pressure compensation of the Z-axis position or the deceleration adjustment of the X / Y axis movement speed within microseconds. This solves the problem of accuracy drift caused by random environmental disturbances such as temperature and humidity changes, and ensures that each cutter head can always maintain high-precision cutting quality in a local range.

[0035] Step 600: Calculate the change in the deviation vector caused by mechanical disturbance during the execution of the local compensation adjustment.

[0036] Step 700: Synchronize the deviation vector change to the adjacent sub-controller nodes in real time, so that the adjacent sub-controller nodes can perform motion path correction according to the deviation vector change.

[0037] Specifically, each sub-controller node will broadcast the deviation vector change information caused by mechanical disturbance during the execution of the local compensation adjustment to the adjacent sub-controller nodes in real time via the high-bandwidth communication bus.

[0038] After receiving the deviation vector change information, adjacent sub-controller nodes incorporate it into their local deformation prediction models for calculation, thereby achieving preventative motion path replanning. For example, when cutter head 1 increases downward pressure to compensate for deformation, cutter head 2, upon receiving this change signal, predicts that the tension in its working area will fluctuate. Therefore, before the actual deformation is transmitted to itself, it proactively updates the input parameters of its local deformation prediction model for preventative path replanning. The technical solution of this invention solves the problem of physical coupling interference during multi-cutter head operation, preventing the compensation action of a single cutter head from triggering a chain reaction of secondary material deformation, and feeding local corrections back to a global coordination mechanism to achieve synchronized action of multiple cutter heads.

[0039] Furthermore, the distributed closed-loop control multi-blade collaborative cutting method proposed in this invention also includes: At the end of the cutting task, each sub-controller node will send back all the deformation data recorded throughout the entire processing process to the master control node; The master control node analyzes and processes the full deformation data using machine learning algorithms, and performs parameterized iteration and training on the initial cutting path map and pre-compensation model based on the analysis results.

[0040] Specifically, each sub-controller node records the full deformation data throughout the entire processing, including time, position, deviation, compensation parameters, and other full-cycle deformation data streams. The master control node invokes machine learning frameworks such as deep neural networks or reinforcement learning algorithms to perform pattern recognition on the aggregated full deformation data, extracting the nonlinear mapping relationship between material deformation patterns and compensation actions to retrain the model.

[0041] In some embodiments of the present invention, step 200 specifically includes: Step 210: Construct a digital twin model of the cut object.

[0042] In this embodiment, a digital twin model of the object to be cut is constructed, creating a digital mirror image in virtual space synchronized with the physical entity of the object. The step of constructing the digital twin model of the object further includes configuring the physical properties and processing environment parameters of the object within the digital twin model, enabling the system to simulate the mechanical tensile force and thermal stress generated by cutting heat on the material as each cutting head moves along its initial path. The physical properties include, but are not limited to, the elastic modulus of various parts of the object to be cut. Poisson's ratio and coefficient of thermal expansion The processing environment parameters include, but are not limited to, preset tension. and ambient temperature .

[0043] Step 220: Simulate the stress field changes generated during the cutting process in virtual space based on the digital twin model.

[0044] For example, in the digital twin model, when any cutting head moves to the path point At that time, based on the digital twin model at the path point The material properties at the location are used to calculate the path points. Stress intensity at Wherein, the stress intensity Represents the path point The instantaneous stress scalar at point . Where the subscript... For the counting variable of the cutting head, used to represent the number of... One cutting blade.

[0045] Step 230: Pre-generate dynamic stress field maps for future path points.

[0046] The dynamic stress field map includes the displacement field of the object being cut during the cutting process. and stress tensor By solving partial differential equations, the displacement field and stress tensor of the object being cut in the global coordinate system can be simulated during the cutting process, thus constructing the dynamic stress field map. Based on the dynamic stress field map, it can be determined which areas will bulge and which areas will stretch during the cutting process according to the original cutting plan.

[0047] Step 240: Perform slicing decomposition on the spatial region of the digital twin model of the object to be cut to obtain several sub-regions.

[0048] Specifically, the processing area of ​​the entire object to be cut is divided into There are 3 adjacent sub-regions, each corresponding to the movement trajectory range of a cutting head, that is, a total of 1,000,000. There are 10 cutting heads, each corresponding to a sub-region. For each cutting head, statistical features within the corresponding sub-region are extracted from the dynamic stress field map as the regional features of that sub-region. For example, the maximum value, average value, and point of maximum stress gradient of the displacement vector within each sub-region are calculated as the regional features of the corresponding sub-region.

[0049] Step 250: Map the regional features of each sub-region to the parameter compensation space to obtain the initial compensation parameter set.

[0050] Step 251: Calculate the... Z-axis height compensation parameters for each sub-region: , in For vector fields of Axial components, The average value function is the Z-axis height compensation parameter. For the first The average vertical displacement predicted within each sub-region.

[0051] Step 252: Calculate the... X / Y axis motion velocity gain for each sub-region: , in, The speed symbol represents For motion speed gain, For the first The predicted strain of the material of the cut object along the material tensile direction within each sub-region.

[0052] Step 250: Calculate the... Local pressure regulation coefficient for each sub-region: , in, The symbol for pressure indicates This is the pressure regulation coefficient. For the first The maximum normal stress predicted within each sub-region. In the processing of flexible materials, stress concentration is a major cause of deformation and tearing. The maximum normal stress... The higher the value, the more unstable the physical state of the region, and the greater the risk of nonlinear deformation of the material. The control law function is a pre-designed empirical function used to control the maximum normal stress. Exceeding the preset critical threshold When, reduce the local pressure adjustment coefficient When the maximum normal stress When the preset critical threshold is exceeded, it indicates that the stress in the corresponding sub-region is too high. It is necessary to reduce the gain of the compensation action to prevent the secondary mechanical disturbance caused by overcompensation from further aggravating the stress concentration.

[0053] Step 254: Use the Z-axis height compensation parameters The X / Y axis motion velocity gain and the local pressure regulation coefficient Construct the initial compensation parameter set: .

[0054] In some embodiments of the present invention, the control law function For exponential buffer functions: , in, This is the reference pressure gain coefficient under standard operating conditions. This refers to the exponential decay coefficient. The pre-configured warning coefficient determines the severity of the system's protective action. Specifically, when the maximum normal stress... Only slightly exceeding the critical threshold At that time, the local pressure adjustment coefficient The stress decreases slowly without affecting processing efficiency; however, when the stress increases explosively, the maximum normal stress... When the local pressure adjustment coefficient increases rapidly, The effect will decrease rapidly, and the executor will enter the system's protection mode, thus allowing reaction time for the subsequent collaborative feedback mechanism.

[0055] In some embodiments of the present invention, the master control node pre-generates a cutting path map of the cutting object and an initial compensation parameter set for each cutter head region. The initial compensation parameter set for each tool head After being encapsulated into structured data packets, these personalized data packets for each sub-region are distributed to each sub-controller via a high-bandwidth bus such as EtherCAT, so that each sub-controller can receive the initial compensation parameter set. Then, it is loaded into the local running memory as the initial setting value for its adaptive PID algorithm. This implementation utilizes the computing power of the master control node to complete complex physical and mathematical calculations before processing, reducing the burden on the edge controller and thus achieving microsecond-level real-time compensation.

[0056] In some embodiments of the present invention, step 400 specifically includes: Step 410: Based on the transient stress wave signal generated by the microcracks inside the material Calculate the first deviation vector: , in These are the weighting coefficients.

[0057] In some embodiments of the present invention, the transient stress wave signal The signal is a high-frequency stress pulse acquired by an acoustic emission sensor mounted on the cutter head of the sub-controller node. When microcracks appear inside the material being cut, the acoustic emission sensor captures the transient stress wave signal. And integrate it over a time window of a specific size to transform it into the first deviation vector. To reflect the first deviation vector Due to its instantaneous characteristics, the integration window is configured as a very short time window, and the system utilizes the first deviation vector. To identify the risk of damage to the object being cut.

[0058] Furthermore, the weighting coefficients It is a Gaussian distribution function centered at the peak of the pulse within a set time window: , Among them, in the above Gaussian distribution function is the standard deviation of the Gaussian distribution, and is a hyperparameter used to adjust the width of the feature extraction window.

[0059] In the technical solution of the embodiments of the present invention, when processing the transient stress wave signal, the weighting coefficients are... Configure as time The transient stress wave signal is implemented using a function rather than a constant. Dynamic filtering is implemented. Specifically, when microcracks occur inside the material of the object being cut, the released energy erupts within an extremely short time. Therefore, at the moment the signal is generated, the weighting coefficients... The weighting coefficients should be configured with higher weights to capture the high-frequency pulse characteristics at that instant; while during the signal attenuation phase, the weighting coefficients should be... It should be attenuated rapidly to ignore the long-tailed, low-frequency interference signals that subsequently arise due to mechanical friction or background ambient noise.

[0060] Step 420: Based on the actual pressure center coordinates during the cutting process Calculate the second deviation vector: , in, These are the predicted pressure center coordinates pre-configured in the cutting path map.

[0061] In the technical solution of the above embodiments, the actual pressure center coordinates and the predicted pressure center coordinates The coordinates are two-dimensional coordinates on the planar coordinate system of the surface of the object being cut.

[0062] In some embodiments, the actual pressure center coordinates The actual pressure center coordinates can be obtained by performing visual analysis and calculations on the real-time images acquired by the vision sensor. In other embodiments, an array of flexible tactile sensors can be arranged on the cutting platform used to support and place the object to be cut to acquire real-time pressure data, and the coordinates of the actual pressure center can be obtained by analyzing this real-time pressure data. .

[0063] Step 430: Based on the actual kerf depth variation rate during the cutting process Calculate the third deviation vector: , in, It is the predicted cut depth variation rate pre-configured in the cutting path map.

[0064] The actual kerf depth variation rate and the predicted rate of change of cut depth The rate of change of the kerf depth over time can be, for example, based on a preset sampling interval. Calculate the actual kerf depth variation rate: , in, and Parting Time and time The kerf depth is obtained by the laser displacement sensor.

[0065] In the above steps, when a sudden increase in the rate of decrease of the kerf depth is detected, it indicates that the material of the object being cut may be tearing.

[0066] In some embodiments of the present invention, step 500 specifically includes: Step 510: Calculate the attenuation coefficient of the X / Y axis motion velocity: , in, The first deviation vector The length of the mold, This is a preset damage sensitivity attenuation constant.

[0067] The technical solution of the above embodiments utilizes the first deviation vector. The amplitude is calculated using a nonlinear mapping function, which serves as the adjustment factor for the current cutting head's forward speed, i.e., the attenuation coefficient. When a high-energy stress wave signal is detected inside the material of the object being cut, the feed rate is actively reduced to prevent the propagation of microcracks, and the movement speed is adjusted to suppress secondary damage to the material structure of the object being cut by mechanical disturbance.

[0068] Step 520: Calculate the incremental compensation value for the Z-axis motion position: , in, The second deviation vector The length of the mold, The second deviation vector The angle between the Z-axis and the positive Z-axis.

[0069] The technical solution of the above embodiment extracts the second deviation vector. The component in the vertical direction, combined with the preset local pressure adjustment coefficient in the cutting path map. Calculate the instantaneous displacement increment used to correct the depth of cutter indentation. This is to compensate for the inconsistency in kerf depth caused by the height fluctuation of the material surface.

[0070] Step 530: Based on the third deviation vector The sign of the time derivative determines the Z-axis compensation amount. The positive and negative directions.

[0071] Step 531: Calculate the third deviation vector using the first-order backward difference operator. The derivative of the change on the time axis: .

[0072] In the technical solutions of the above embodiments, The value represents whether the kerf depth deviation is increasing or decreasing. When When this occurs, it indicates that the deviation is worsening, meaning the rate of change in depth is drifting away from the preset value. And when... When the deviation is converging, it means that the system is regressing towards the preset path.

[0073] Step 532: Construct the decision mapping function: , in, The function is a symbolic function, used for determination. still ,Right now: when hour, .

[0074] when hour, .

[0075] Step 533: Determine the Z-axis incremental compensation amount based on the decision mapping function. The plus or minus sign.

[0076] when When this is determined to be deviation expansion, the control command execution mechanism performs an upward movement of the Z-axis to reduce the cutting depth. When the deviation convergence is determined, the control command execution mechanism maintains or executes the Z-axis downward pressing action to maintain or increase the cutting depth.

[0077] Step 540: The incremental compensation value based on the Z-axis motion position. The attenuation coefficient of the X / Y axis motion speed, and the controller drives the actuator to adjust the parameters of the tool head position and motion speed.

[0078] In some embodiments of the present invention, step 600 specifically includes: Step 610: Construct the deviation vector sequence between the current time and the previous sampling time: .

[0079] Step 620: Calculate the time-domain difference change of the deviation vector sequence: .

[0080] Step 630: Calculate the scalar magnitude of the deviation disturbance caused by the mechanical disturbance: .

[0081] In the technical solution of the above embodiments, the time-domain difference change of the deviation vector sequence The magnitude of the time-domain differential change of the deviation vector sequence reflects the rate of disturbance to the global stress field caused by local correction actions. Used to characterize the severity of a disturbance.

[0082] The deviation disturbance intensity scalar For the time-domain difference change The scalar index obtained after normation processing can be used to assess the risk of cascading effects of local compensation actions on adjacent tool tip regions. The deviation disturbance intensity scalar... The larger the value, the more significant the mechanical disturbance caused by the current tool head's correction action, requiring the triggering of a higher-level collaborative early warning mechanism.

[0083] In some embodiments of the present invention, step 700 specifically includes: Step 710: Execute differential information broadcast based on the EtherCAT protocol so that adjacent sub-controller nodes can broadcast the time-domain differential change. and / or the deviation disturbance intensity scalar Perform local forecast data updates.

[0084] In this invention, the term "adjacent sub-controller node" refers to a sub-controller node arranged on adjacent tool heads. In this embodiment, the broadcast mechanism of a high-bandwidth communication bus is used to transmit the time-domain differential change calculated in step 600. and / or the deviation disturbance intensity scalar It is encapsulated into a real-time data frame and synchronously pushed to each sub-controller node that is spatially adjacent.

[0085] Step 720: Neighboring sub-controller nodes recalculate their own compensation parameter sets based on the updated local prediction data.

[0086] The local prediction data is generated by the local deformation prediction model on the adjacent sub-controller nodes based on the time-domain difference change. and / or the deviation disturbance intensity scalar The generated dynamic stress field map and compensation parameter set are then incorporated into the input feature space of the localized local deformation prediction model by adjacent sub-controller nodes after receiving the broadcast information. The dynamic stress field map and compensation parameter set in their preset path features are updated through a weight adjustment algorithm.

[0087] like Figure 2 As shown, a second aspect of the present invention proposes a distributed closed-loop controlled multi-blade collaborative cutting device, comprising: The master node is used to execute global task decomposition and pre-compensation planning, and to simulate and generate the initial cutting path map and the initial compensation parameter set for each cutter head region based on the material physical properties of the cutting object. The distributed sub-controller node array is deployed on multiple interconnected cutting heads. Each sub-controller node corresponds to one cutting head. Each sub-controller node integrates a localized deformation prediction model, which is used to receive initial parameters issued by the master node and perform local closed-loop compensation based on real-time sensing data. A multi-dimensional sensor array, integrated on each of the sub-controller nodes, is used to collect sound, light, and pressure information of the object being cut during the cutting process; A high-bandwidth communication bus is connected to the master control node and the sub-controller nodes to enable real-time command issuance and broadcast synchronization of deviation vector information between sub-controller nodes. The multi-blade collaborative cutting device is configured to implement a multi-blade collaborative cutting method with distributed closed-loop control as described in any of the first aspects of the present invention.

[0088] In the technical solution of this invention, the multi-head collaborative cutting device uses a high-bandwidth communication bus as the data carrier. The master control node generates and sends cutting path maps and compensation parameter sets to each sub-controller node. These maps and parameters provide prior knowledge and benchmark references to the cutting process, transforming complex physical deformation problems into preset parameters for geometric paths, thus reducing the search space and computational difficulty for subsequent real-time control. The specific real-time trajectory calculation is handled by each sub-controller node, avoiding the communication delay caused by large-scale sensor data transmission back to the master control node and subsequent command return.

[0089] In some embodiments of the present invention, after the cutting task is completed, each sub-controller node uploads the recorded full-cycle deformation data stream, including time, position, deviation, and compensation actions, to the master control node. The master control node then uses machine learning algorithms to optimize and iterate the local deformation prediction model based on this full-cycle deformation data.

[0090] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0091] As described above, these embodiments of the present invention do not exhaustively cover all details, nor do they limit the invention to the specific embodiments described. Clearly, many modifications and variations can be made based on the above description. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to effectively utilize the invention and its modifications. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A distributed closed-loop control method for multi-blade collaborative cutting, characterized in that, include: Generate the cutting path map of the cutting object and the initial compensation parameter set for each cutter head region; The cutting path map and initial compensation parameter set are distributed in real time to the sub-controller nodes deployed on multiple cutting heads; Real-time acquisition of multi-dimensional sensing features during the cutting process; Calculate the deviation vector between the multi-dimensional perception features at the current moment and the preset path features; The Z-axis position and X / Y-axis speed of the corresponding tool head are locally compensated and adjusted according to the deviation vector. Calculate the change in the deviation vector caused by mechanical disturbance during the execution of the local compensation adjustment; The deviation vector change is synchronized to the adjacent sub-controller nodes in real time, so that the adjacent sub-controller nodes can perform motion path correction based on the deviation vector change.

2. The distributed closed-loop control multi-blade collaborative cutting method according to claim 1, characterized in that, The specific steps for generating the initial cutting path map and the initial compensation parameter set for each cutter head region include: Construct a digital twin model of the object to be cut; Based on the aforementioned digital twin model, the stress field changes generated during the cutting process are simulated in virtual space; Pre-generate dynamic stress field maps for future path points; The spatial region of the digital twin model of the object to be cut is decomposed into several sub-regions by slicing. The initial compensation parameter set is obtained by mapping the regional features of each sub-region to the parameter compensation space.

3. The distributed closed-loop control multi-head collaborative cutting method according to claim 2, characterized in that, The specific steps of mapping the regional features of each sub-region to the parameter compensation space to obtain the initial compensation parameter set include: Calculate the first Z-axis height compensation parameters for each sub-region: ， in For vector fields of Axial components, It is a function of average value; Calculate the first X / Y axis motion velocity gain for each sub-region: ， in For the first The predicted strain of the material of the cut object along the tensile direction within each sub-region; Calculate the first Local pressure regulation coefficient for each sub-region: ， in For the first The predicted maximum normal stress in each sub-region For pre-configured control law functions; Using the Z-axis height compensation parameters The X / Y axis motion velocity gain and the local pressure regulation coefficient Construct the initial compensation parameter set: 。 4. The distributed closed-loop control multi-head collaborative cutting method according to claim 3, characterized in that, The control law function For an exponential buffer function, calculate the th Local pressure regulation coefficient of each sub-region The specific steps are as follows: ， in This is the reference pressure gain coefficient under standard operating conditions. The pre-configured exponential decay coefficient.

5. The distributed closed-loop control multi-head collaborative cutting method according to claim 1, characterized in that, The specific steps for calculating the deviation vector between the multi-dimensional perceptual features at the current moment and the preset path features include: Based on the transient stress wave signal generated by microcracks inside the material Calculate the first deviation vector: ， in These are the weighting coefficients; Based on the actual pressure center coordinates during the cutting process Calculate the second deviation vector: ， in These are the predicted pressure center coordinates pre-configured in the cutting path map; Based on the actual kerf depth change rate during the cutting process Calculate the third deviation vector: ， in It is the predicted cut depth variation rate pre-configured in the cutting path map.

6. The distributed closed-loop control multi-head collaborative cutting method according to claim 5, characterized in that, The weighting coefficient It is a Gaussian distribution function centered at the peak of the pulse within a set time window: ， Among them, in the above Gaussian distribution function is the standard deviation of the Gaussian distribution, and is a hyperparameter used to adjust the width of the feature extraction window.

7. The distributed closed-loop control multi-head collaborative cutting method according to claim 5, characterized in that, The steps for locally compensating and adjusting the Z-axis position and X / Y-axis speed of the tool head based on the deviation vector specifically include: Calculate the attenuation coefficient of the X / Y axis motion velocity: ， in, The first deviation vector The length of the mold, This is a preset damage sensitivity attenuation constant; Calculate the incremental compensation value for the Z-axis motion position: ， in, The second deviation vector The length of the mold, The second deviation vector The angle with the positive Z-axis; The Z-axis incremental compensation amount is determined based on the decision mapping function. The plus or minus sign; The incremental compensation value based on the Z-axis motion position The attenuation coefficient of the X / Y axis motion speed, and the controller drives the actuator to adjust the parameters of the tool head position and motion speed.

8. The distributed closed-loop control multi-head collaborative cutting method according to claim 5, characterized in that, The steps for calculating the change in the deviation vector caused by mechanical disturbance during the execution of the local compensation adjustment specifically include: Construct the deviation vector sequence between the current time and the previous sampling time: ； Calculate the time-domain difference change of the deviation vector sequence: ； Calculate the scalar value of the deviation disturbance intensity caused by mechanical disturbance: 。 9. The distributed closed-loop control multi-head collaborative cutting method according to claim 5, characterized in that, The step of synchronizing the deviation vector change to adjacent sub-controller nodes in real time, so that the adjacent sub-controller nodes can perform motion path correction based on the deviation vector change, specifically includes: Perform differential information broadcasting based on the EtherCAT protocol so that adjacent sub-controller nodes can adjust according to the time-domain differential change. and / or the deviation disturbance intensity scalar Perform local forecast data updates; Neighboring sub-controller nodes recalculate their own compensation parameter sets based on the updated local prediction data.

10. A distributed closed-loop controlled multi-head collaborative cutting device, characterized in that, include: The master node is used to execute global task decomposition and pre-compensation planning, and to simulate and generate the initial cutting path map and the initial compensation parameter set for each cutter head region based on the material physical properties of the cutting object. The distributed sub-controller node array is deployed on multiple interconnected cutting heads. Each sub-controller node corresponds to one cutting head. Each sub-controller node integrates a localized deformation prediction model, which is used to receive initial parameters issued by the master node and perform local closed-loop compensation based on real-time sensing data. A multi-dimensional sensor array, integrated on each of the sub-controller nodes, is used to collect sound, light, and pressure information of the object being cut during the cutting process; A high-bandwidth communication bus is connected to the master control node and the sub-controller nodes to enable real-time command issuance and broadcast synchronization of deviation vector information between sub-controller nodes. The multi-head collaborative cutting device is configured to implement the distributed closed-loop control multi-head collaborative cutting method as described in any one of claims 1-9.

Images