Control system for improving numerical control machining precision of bionic unmanned aerial vehicle wing

Abstract

The invention discloses a control system for improving the numerical control machining precision of a bionic unmanned aerial vehicle wing, and relates to the professional technical field of numerical control machining, and the control system specifically comprises a multi-dimensional sensing module, a digital twin model construction module and a numerical control machining precision improvement module, the multi-dimensional sensing module acquires geometric and material states and dynamic response data through a high-precision sensor cluster and realizes space-time consistency, the digital twin model building module comprises geometric modeling and physical attribute modeling, and is used for predicting the service life of a tool by combining an Archard wear model based on laser scanning point cloud and CAD (Computer Aided Design) model registration compensation pose deviation. And meanwhile, a five-axis machine tool multi-body dynamic model is established, a numerical control machining precision improving module dynamically divides sub-routes, establishes an error prediction model, corrects a machining path in real time, regulates and controls monitoring frequency according to geometric errors, thermal errors and material damage errors, triggers a cooling system to enhance or bypass a damage area, and guarantees machining precision.

Description

Technical Field

[0001] The present invention belongs to the technical field of numerical control machining, and in particular relates to a control system for improving the numerical control machining accuracy of a bionic unmanned aerial vehicle (UAV) wing. Background Art

[0002] Bionic drones are designed to mimic the flight characteristics of natural organisms, such as the agile flight of birds and the high maneuverability of insects, to achieve more efficient, stealthy, and adaptable flight capabilities in complex environments. As a key component of drone flight performance, the shape, structure, and surface quality of the wing directly impact aerodynamic performance, including lift, drag, and stability. However, during CNC machining, multiple factors, including geometric error, thermal error, and material damage error, interact and affect machining accuracy. Geometric error primarily stems from manufacturing and assembly errors within the machine tool itself, as well as wear and tear of moving parts. Thermal error is caused by thermal deformation of machine tool components due to factors such as cutting heat and frictional heat during machining. Material damage error is closely related to material properties, machining parameters, and the mechanical and thermal behavior during machining. Traditional CNC machining systems struggle to fully and in real time perceive and compensate for these multi-dimensional errors, hindering further improvements in machining accuracy.

[0003] Therefore, there is an urgent need for a control system that can improve the CNC machining accuracy of bionic UAV wings. Through the deep integration of multi-dimensional perception, digital twin model construction and adaptive control, a precision-efficiency-reliability processing solution is provided for high-value complex components such as bionic wings. Summary of the Invention

[0004] The purpose of the present invention is to provide a control system for improving the CNC machining accuracy of bionic UAV wings, which is used to solve the technical problem that the existing technology is difficult to adapt to the dynamic compensation and prediction requirements during the machining process.

[0005] In order to achieve the above object, the present invention adopts the following technical solutions:

[0006] A control system for improving the numerical control machining accuracy of bionic UAV wings, comprising:

[0007] Multi-dimensional perception module, which acquires geometry, material state and dynamic response data through a high-precision sensor cluster and achieves spatiotemporal consistency;

[0008] The digital twin model construction module includes geometric modeling and physical property modeling, compensating for pose deviation based on laser scanning point cloud and CAD model registration, predicting tool life in combination with the Archard wear model, and establishing a multi-body dynamics model for five-axis machine tools.

[0009] The CNC machining precision improvement module dynamically divides sub-routes and establishes an error prediction model to correct the machining path in real time.

[0010] Furthermore, a high-precision sensor cluster is used to obtain geometry, material state, and dynamic response data, and achieve spatiotemporal consistency. The specific method is as follows:

[0011] Geometric perception layer: A 3D line laser scanner with a specific resolution and scanning rate is installed along the X / Y axis guide rails of the CNC machine tool. Bilateral filtering is used to remove outlier noise in the point cloud and cover a specific range of the processing area.

[0012] Material state perception layer: N acoustic emission sensors with a fixed bandwidth are arranged on the base of the workpiece fixture. Wavelet threshold denoising and triangulation positioning are combined to capture carbon fiber fracture signals. An infrared thermal imager with a specific sensitivity is configured to monitor the temperature field at a specific frame rate.

[0013] Dynamic response sensing layer: A six-dimensional force sensor with specific range and error is integrated into the spindle-tool interface, and a high-frequency vibration sensor with a sensitivity of H is used in the tool overhang area.

[0014] Furthermore, the pose deviation is compensated based on the registration of the laser scanning point cloud and the CAD model. The specific method is as follows:

[0015] Based on feature point matching, the laser scanning point cloud and the CAD model are registered by ICP, and the weighted iterative closest point algorithm is used to calculate the pose deviation compensation ΔP = [Δx, Δy, Δz, Δθ x , Δθ y , Δθ z ,], where Δx, Δy, Δz are translation deviations, Δθ x , Δθ y , Δθ z For rotation deviation, the motion compensation is calculated by kinematic model The deviation components are converted to obtain, where ΔX, ΔY and ΔZ represent the motion compensation required to be applied to the X / Y / Z axis of the machine tool, respectively, and R x , R y With R z Respectively represent the curvature radius components of the tool in the X / Y / Z axis directions.

[0016] Furthermore, the tool life is predicted by combining the Archard wear model. The specific method is as follows:

[0017] For the laminated structure of carbon fiber composite materials, the orthotropic elastic matrix is defined, and the Hashin criterion or Tsai-Wu criterion is used to predict the critical conditions of fiber fracture and matrix cracking. The temperature field distribution model is established based on Fourier's law to simulate the influence of cutting heat on the material. The Archard wear model is established through the relationship between wear volume, sliding distance and normal force. represents the Archard wear model, where V represents the tool wear volume, H represents the hardness of the workpiece material, K represents the dimensionless wear coefficient calibrated by experiments, L represents the relative sliding distance between the tool and the workpiece, which is calculated by multiplying the cutting speed and the machining time, and F represents the normal force at the contact surface between the tool and the workpiece;

[0018] The sensor data is sampled every 5ms, the instantaneous wear volume increment is calculated, and the total wear volume is obtained by summing up all the instantaneous wear volume increments. The critical failure volume is determined in advance through experiments. When the total wear volume is greater than or equal to the critical failure volume, the tool change command is triggered. The ratio of the instantaneous wear volume increment to the sampling frequency is recorded as the wear rate. The critical failure volume is subtracted from the current total wear volume to obtain the residual wear volume. The ratio of the residual wear volume to the current wear rate is the predicted remaining processing time.

[0019] Furthermore, a multi-body dynamics model of a five-axis machine tool is established. The specific method is as follows:

[0020] A multi-body dynamics model of a five-axis machine tool is established in Simulink. The matrix forms of mass M, damping C, and stiffness K are obtained by frequency response function fitting. The spindle-guideway-screw coupled vibration equation is shown as follows:

[0021]

[0022] Among them, M is the mass matrix, which represents the inertia of each moving part, C is the damping matrix, which represents the energy dissipation of the system, K is the stiffness matrix, which represents the elastic restoring force of the system, and q is the displacement of each degree of freedom of the machine tool, which is expressed as x, y, z represent the linear displacement of the tool in the Cartesian coordinate system, θ A is the angular displacement around axis A, θ C is the angular displacement around the C axis, is the velocity vector containing all degrees of freedom of the machine tool, expressed as Where v represents velocity, ω represents angular velocity,

[0023] is the acceleration vector containing all degrees of freedom of the machine tool, expressed as Where a represents acceleration, α represents angular acceleration, and FC represents cutting force vector.

[0024] Furthermore, sub-routes can be divided dynamically. The specific method is as follows:

[0025] In the machining path planning stage, the curvature gradient of each path point is calculated in real time through the second-order derivative calculation of the point cloud. The maximum allowable curvature change and the allowable material removal rate are determined through machining tests. The feed step length and feed speed are determined according to the initial setting of the process parameters of the CNC machining machine tool. The product of the feed step length and feed speed is used to express the instantaneous material removal rate. The formula is used represents the sub-route division rule, where maxk represents the maximum allowable curvature change, Δk represents the curvature change rate, and Q p Indicates the allowable material removal rate, a p Indicates the feed step length, v p Indicates feed speed, L i Indicates the length of the i-th sub-route, that is, L is equal to and The minimum value in the original continuous path is divided into L i Interval subroutes.

[0026] Furthermore, an error prediction model is established. The specific method is as follows:

[0027] After the sub-route is determined, the monitoring frequency is adjusted according to the error prediction model, using the formula

[0028] Represents the error prediction model, where R i represents the curvature radius of the sub-route calculated by the path geometry, k s represents the system equivalent stiffness determined by the hammering method, ΔT represents the processing temperature rise measured in real time by the infrared thermal imager, and D f Represents the acoustic emission energy, which is obtained by wavelet energy analysis of the acoustic emission signal.

[0029] Furthermore, the monitoring frequency is adjusted according to the error prediction model. The specific method is as follows:

[0030] Frequency control function Control monitoring frequency, where f b Indicates the reference frequency, that is, the minimum effective sampling requirement of the sensor, f m represents the maximum frequency, i.e. the sensor hardware limit, τ is the accuracy standard, and γ is the precision factor.

[0031] Furthermore, the processing path is corrected in real time. The specific method is as follows:

[0032] Set the geometric error threshold to 0.5τ. When it is greater than or equal to 0.5τ, use the formula P E =P N +ΔP performs reverse correction on the tool path, where PN Represents the coordinates of the next tool path point originally generated, ΔP represents the path correction vector, and the x-direction component is equal to The y-direction component is equal to The z-direction component is equal to

[0033] where R x , R y , R z Directional components of the radius of curvature on the x-axis, y-axis, and z-axis;

[0034] when When it is less than 0.5τ, it is considered normal and no processing is performed;

[0035] When the monitored temperature is greater than or equal to the temperature safety threshold, the cooling system is triggered to be enhanced, using the formula Dynamically adjust the coolant flow rate according to the temperature rise, where LE b represents the standard coolant flow rate, ε is the temperature change factor, T is the monitored instantaneous temperature, maxT is the temperature safety threshold, and ΔT is the instantaneous processing temperature rise at the monitoring time point, that is, the temperature difference between the current monitoring time point and the previous monitoring time point;

[0036] When the monitored temperature is lower than the temperature safety threshold, it is considered normal and no action is taken;

[0037] When the acoustic emission energy D f When the fiber damage threshold D is greater than or equal to the fiber damage threshold, the next sub-route is judged to be in the damage area and the sub-route in the damage area is automatically bypassed. f When it is less than the fiber damage threshold D, the next sub-route is judged to be in the normal area and no processing is performed.

[0038] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are:

[0039] 1. This invention uses a high-precision sensor cluster to comprehensively and in real time acquire in-process geometry, material state, and dynamic response data, achieving spatiotemporal consistency and effectively avoiding processing errors caused by data deviations. The geometric perception layer accurately acquires and removes noise from the blank surface point cloud data to ensure processing accuracy. The material state perception layer captures carbon fiber fracture signals and monitors the temperature field to promptly detect material damage and thermal problems. The dynamic response perception layer collects cutting force, torque vector, and chatter frequency in real time to help dynamically adjust processing parameters, ensure a stable processing process, and improve product quality.

[0040] 2. This invention aligns the laser scanning point cloud with the CAD model and compensates for posture deviations, converting the deviations into motion compensation for each axis. This enables high-precision machining path planning and adjustment, significantly improving machining accuracy. Physical property modeling is used to predict the critical conditions for fiber breakage and matrix cracking, simulates the effects of cutting heat, and combines the Archard wear model to predict tool life. This allows for advance planning of tool changes, avoiding machining errors caused by excessive tool wear, and improving machining efficiency and quality. Furthermore, a multi-body dynamics model of a five-axis machine tool is established to accurately simulate the dynamic characteristics of the machine tool, providing a basis for optimizing machining parameters.

[0041] 3. The present invention dynamically divides sub-routes based on the curvature gradient and material removal rate constraints during the processing path planning stage, effectively avoiding processing errors caused by sudden changes in path curvature or uneven material removal, and improving processing accuracy. According to the geometric characteristics of the sub-routes, system stiffness, processing temperature rise and acoustic emission energy, the monitoring frequency is regulated by the error prediction model, resources are accurately allocated, and monitoring efficiency and accuracy are improved. At the same time, the processing path is corrected in real time based on the geometric error, thermal error and material damage error to ensure processing accuracy. By setting the geometric error threshold, temperature safety threshold and fiber damage threshold, intelligent decision-making and adaptive adjustment of the processing process are achieved, ensuring processing accuracy and stability, and improving processing efficiency and product quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0043] Figure 1 Shown is a control system module diagram for improving the CNC machining accuracy of bionic UAV wings;

[0044] Figure 2 A control system workflow diagram for improving the CNC machining accuracy of bionic UAV wings is shown;

[0045] Figure 3 The main working steps of the CNC machining precision improvement module are shown. DETAILED DESCRIPTION

[0046] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0047] like Figure 1 、 Figure 2 and Figure 3 As shown, a control system for improving the CNC machining accuracy of bionic UAV wings specifically includes the following contents:

[0048] The multi-dimensional perception module acquires geometry, material state, and dynamic response data through a high-precision sensor cluster, achieving temporal and spatial consistency. To meet the processing requirements of bionic wings, the following high-precision sensor clusters are configured:

[0049] Geometric perception layer: A 3D line laser scanner with a resolution of 5μm and a scanning rate of 10kHz is installed along the X / Y axis guide rails of the CNC machine tool to obtain real-time point cloud data of the blank surface. Bilateral filtering is used to remove outlier noise in the point cloud data, covering the processing area within ±200mm.

[0050] Material status perception layer: Four acoustic emission sensors with a bandwidth of 50kHz to 400kHz are evenly arranged on the base of the workpiece fixture. Wavelet threshold denoising is used to retain high-frequency damage characteristics, and carbon fiber fracture signals are captured through triangulation positioning. The test benchmark is an infrared thermal imager that can detect a temperature difference of 0.03°C at an ambient temperature of 30°C, and the temperature field in the processing area is monitored at a frame rate of 100Hz.

[0051] Dynamic response perception layer: A six-dimensional force sensor with a range of ±500N and a maximum error of no more than ±2.5N is integrated at the interface between the CNC machine tool spindle and the tool to collect the cutting force / torque vector in real time. A high-frequency vibration sensor with a sensitivity of 10mV / g deployed at the tool overhang is used to monitor the chatter frequency.

[0052] By adopting the IEEE 1588PTP precision time protocol, the timestamp deviation of each sensor data is ensured to be less than 1μs. The degree of freedom transformation relationship between the sensor coordinate system and the machine tool coordinate system is established through the nine-point calibration method to achieve the spatiotemporal consistency of multi-sensor data.

[0053] The digital twin model construction module includes geometric modeling and physical property modeling, compensates for posture deviation based on laser scanning point cloud and CAD model registration, predicts tool life in combination with the Archard wear model, and establishes a five-axis machine tool multi-body dynamics model.

[0054] Geometric modeling: Based on feature point matching, the laser scanning point cloud and the CAD model are registered by ICP, and the weighted iterative closest point algorithm is used to calculate the pose deviation compensation ΔP = [Δx, Δy, Δz, Δθ x , Δθ y , Δθ z ,], where Δx, Δy, Δz are translation deviations, Δθ x, Δθ y , Δθ z The deviation is the rotation deviation. Based on the kinematic model of the machine tool, the deviation is converted into the motion compensation of each axis. The motion compensation is obtained by converting the deviation component through the kinematic model:

[0055]

[0056] Where ΔX, ΔY and ΔZ represent the motion compensation required to be applied to the X / Y / Z axis of the machine tool, respectively. x , R y With R z They represent the curvature radius components of the tool in the X / Y / Z axis directions respectively. The curvature radius components are obtained by the sum of the preset nominal radius and the wear increment.

[0057] Physical property modeling: For the laminated structure of carbon fiber composites, an orthotropic elastic matrix is defined. The Hashin criterion or Tsai-Wu criterion is used to predict the critical conditions for fiber fracture and matrix cracking, such as the fiber direction tensile failure criterion. A temperature field distribution model is established based on Fourier's law to simulate the effect of cutting heat on the material. The Archard wear model is established based on the relationship between wear volume, sliding distance, and normal force. The wear volume estimate is updated online by combining the high-frequency energy of the acoustic emission signal with the cutting force fluctuation. Tool life is predicted based on the Archard wear model. The specific formula of the Archard wear model is as follows:

[0058]

[0059] Among them, V represents the tool wear volume, H represents the hardness of the workpiece material, K represents the dimensionless wear coefficient calibrated by experiments, L represents the relative sliding distance between the tool and the workpiece, which is calculated by the product of the cutting speed and the processing time, and F represents the normal force at the contact surface between the tool and the workpiece. The cutting force vector is directly measured by a six-dimensional force sensor, and the normal component is extracted through coordinate system transformation.

[0060] The sensor data is sampled every 5ms, the instantaneous wear volume increment is calculated, and the total wear volume is obtained by summing up all the instantaneous wear volume increments. The critical failure volume is determined in advance through experiments. When the total wear volume is greater than or equal to the critical failure volume, the tool change command is triggered. The ratio of the instantaneous wear volume increment to the sampling frequency is recorded as the wear rate. The critical failure volume is subtracted from the current total wear volume to obtain the residual wear volume. The ratio of the residual wear volume to the current wear rate is the predicted remaining processing time.

[0061] A multi-body dynamics model of a five-axis machine tool was established in Simulink. Based on the Lagrange equation, the kinetic energy, potential energy, and dissipated energy of each moving component of the machine tool (spindle, guide rail, lead screw, etc.) were modeled. Pulse excitation was applied to each degree of freedom of the machine tool. The response signal was collected by the accelerometer. The matrix form of mass M, damping C, and stiffness K was obtained by frequency response function fitting. The spindle-guide rail-lead screw coupled vibration equation is shown as follows:

[0062]

[0063] Among them, M is the mass matrix, which represents the inertia of each moving part, C is the damping matrix, which represents the energy dissipation of the system, K is the stiffness matrix, which represents the elastic restoring force of the system, and q is the displacement of each degree of freedom of the machine tool, which is expressed as x, y, z represent the linear displacement of the tool in the Cartesian coordinate system, θ A is the angular displacement around the A axis (tilt axis), θ C is the angular displacement around the C axis (rotation axis), is the velocity vector containing all degrees of freedom of the machine tool, expressed as Where v represents velocity, ω represents angular velocity, is the acceleration vector containing all degrees of freedom of the machine tool, expressed as Where a represents acceleration, α represents angular acceleration, and FC represents the cutting force vector, which is obtained from the force sensor data during the machining process.

[0064] The SPH (smoothed particle hydrodynamics) method is used to simulate the carbon fiber delamination and resin removal process, with a particle diameter of 50μm and a time step of 1μs. The thermal-mechanical coupling field is integrated to solve the instantaneous cutting temperature distribution. The sensor data is injected into the twin model in real time through the OPCUA protocol. The refresh cycle is ≤10ms, and a data-driven event trigger is defined: when the force sensor detects a cutting force >300N or the thermal imager displays a temperature >80 degrees, the current control instruction is frozen, the processing status is saved, and high-precision measurement equipment (such as a laser interferometer) is called to recalibrate the tool position and workpiece geometry, and update the initial conditions and boundary parameters of the digital twin model.

[0065] The CNC machining precision improvement module dynamically divides sub-routes and establishes an error prediction model to correct the machining path in real time. It also adjusts the monitoring frequency based on geometric error, thermal error, and material damage error, triggering the cooling system to enhance or bypass damaged areas to ensure machining accuracy.

[0066] During the machining path planning stage, the curvature gradient of each path point is calculated in real time through the second-order derivative calculation of the point cloud. The maximum allowable curvature change and allowable material removal rate are determined through machining tests. The feed step length and feed speed are determined according to the initial setting of the process parameters of the CNC machining machine tool. The product of the feed step length and feed speed is used to express the instantaneous material removal rate. The sub-route division rules are established by combining the curvature gradient and material removal rate constraints. The specific formula is as follows:

[0067]

[0068] Among them, maxk represents the maximum allowable curvature change, Δk represents the curvature change rate, Q p Indicates the allowable material removal rate, a p Indicates the feed step length, v p Indicates feed speed, L i Indicates the length of the i-th sub-route, that is, L is equal to and The minimum value in the original continuous path is divided into L i Interval subroutes.

[0069] After the sub-route is determined, the monitoring frequency is adjusted according to the error prediction model. The specific formula of the error prediction model is as follows:

[0070]

[0071] Among them, R i represents the curvature radius of the sub-route calculated by the path geometry, k s represents the system equivalent stiffness determined by the hammering method, ΔT represents the processing temperature rise measured in real time by the infrared thermal imager, and D f Represents the acoustic emission energy, which is obtained by wavelet energy analysis of the acoustic emission signal and is controlled by the frequency control function Control monitoring frequency, where f b Indicates the reference frequency, that is, the minimum effective sampling requirement of the sensor, f m represents the maximum frequency, i.e., the sensor hardware limit, τ is the accuracy standard, and γ is the precision factor. In this embodiment, τ is determined to be 5 μm and γ is determined to be 1.5.

[0072] Set the geometric error threshold to 0.5τ. When it is greater than or equal to 0.5τ, it is determined that the tool path is reversely corrected. The specific formula is:

[0073] P E =P N +ΔP;

[0074] Among them, P NRepresents the coordinates of the next tool path point originally generated, ΔP represents the path correction vector, and the x-direction component is equal to The y-direction component is equal to

[0075] The z-direction component is equal to where R x , R y , R z The directional components of the radius of curvature on the x-axis, y-axis, and z-axis.

[0076] when When it is less than 0.5τ, it is judged as normal and no processing is performed.

[0077] When the monitored temperature is greater than or equal to the temperature safety threshold (set to 80 degrees Celsius in this embodiment), the cooling system is triggered to enhance and dynamically adjust the coolant flow rate according to the temperature rise. The specific formula is as follows:

[0078]

[0079] Among them, LE b represents the standard coolant flow rate, ε is the temperature change factor, T is the monitored instantaneous temperature, maxT is the temperature safety threshold, and ΔT is the instantaneous processing temperature rise at the monitoring time point, that is, the temperature difference between the current monitoring time point and the previous monitoring time point. In this embodiment, ε is set to 0.3.

[0080] When the monitored temperature is lower than the temperature safety threshold, it is considered normal and no action is taken.

[0081] When the acoustic emission energy D f When the fiber damage threshold D is greater than or equal to the fiber damage threshold, the next sub-route is judged to be in the damage area and the sub-route in the damage area is automatically bypassed. f When it is less than the fiber damage threshold D, the next sub-route is judged to be in the normal area and no processing is performed.

[0082] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with the technical field, within the technical scope disclosed by the present invention, who makes equivalent replacements or changes based on the technical solutions and inventive concepts of the present invention, should be covered by the scope of protection of the present invention.

[0083] The preferred embodiments of the present invention disclosed above are intended only to help illustrate the present invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the present invention to specific embodiments. Obviously, many modifications and variations are possible based on the contents of this specification. These embodiments are selected and described in detail in this specification to better explain the principles and practical applications of the present invention, thereby enabling those skilled in the art to better understand and utilize the present invention. The present invention is limited only by the claims and their full scope and equivalents.

Claims

1. A control system for improving the numerical control machining accuracy of bionic UAV wings, characterized in that: include: Multi-dimensional perception module, which acquires geometry, material state and dynamic response data through a high-precision sensor cluster and achieves spatiotemporal consistency; The digital twin model construction module includes geometric modeling and physical property modeling, compensating for pose deviation based on laser scanning point cloud and CAD model registration, predicting tool life in combination with the Archard wear model, and establishing a multi-body dynamics model for five-axis machine tools. The CNC machining precision improvement module dynamically divides sub-routes and establishes an error prediction model to correct the machining path in real time.

2. A control system for improving the numerical control machining accuracy of bionic drone wings according to claim 1, characterized in that: The geometry, material state, and dynamic response data are acquired through a high-precision sensor cluster, and are spatially and temporally consistent. The specific methods are: Geometric perception layer: A 3D line laser scanner with a specific resolution and scanning rate is installed along the X / Y axis guide rails of the CNC machine tool. Bilateral filtering is used to remove outlier noise in the point cloud and cover a specific range of the processing area. Material state perception layer: N acoustic emission sensors with a fixed bandwidth are arranged on the base of the workpiece fixture. Wavelet threshold denoising and triangulation positioning are combined to capture carbon fiber fracture signals. An infrared thermal imager with a specific sensitivity is configured to monitor the temperature field at a specific frame rate. Dynamic response sensing layer: A six-dimensional force sensor with specific range and error is integrated into the spindle-tool interface, and a high-frequency vibration sensor with a sensitivity of H is used in the tool overhang area.

3. A control system for improving the numerical control machining accuracy of bionic drone wings according to claim 1, characterized in that: Compensation for pose deviation based on registration of laser scanning point cloud and CAD model. The specific method is as follows: Based on feature point matching, the laser scanning point cloud and the CAD model are registered by ICP, and the weighted iterative closest point algorithm is used to calculate the pose deviation compensation ΔP = [Δx, Δy, Δz, Δθ x , Δθ y , Δθ z ,], where Δx, Δy, Δz are translation deviations, Δθ x , Δθ y , Δθ z For rotation deviation, the motion compensation is calculated by kinematic model The deviation components are converted to obtain, where ΔX, ΔY and ΔZ represent the motion compensation required to be applied to the X / Y / Z axis of the machine tool, respectively, and R x , R y With R z Respectively represent the curvature radius components of the tool in the X / Y / Z axis directions.

4. A control system for improving the numerical control machining accuracy of bionic drone wings according to claim 1, characterized in that: The tool life is predicted by combining the Archard wear model. The specific method is as follows: For the laminated structure of carbon fiber composite materials, the orthotropic elastic matrix is defined, and the Hashin criterion or Tsai-Wu criterion is used to predict the critical conditions of fiber fracture and matrix cracking. The temperature field distribution model is established based on Fourier's law to simulate the influence of cutting heat on the material. The Archard wear model is established through the relationship between wear volume, sliding distance and normal force. represents the Archard wear model, where V represents the tool wear volume, H represents the hardness of the workpiece material, K represents the dimensionless wear coefficient calibrated by experiments, L represents the relative sliding distance between the tool and the workpiece, which is calculated by multiplying the cutting speed and the machining time, and F represents the normal force at the contact surface between the tool and the workpiece; The sensor data is sampled every 5ms, the instantaneous wear volume increment is calculated, and the total wear volume is obtained by summing up all the instantaneous wear volume increments. The critical failure volume is determined in advance through experiments. When the total wear volume is greater than or equal to the critical failure volume, the tool change command is triggered. The ratio of the instantaneous wear volume increment to the sampling frequency is recorded as the wear rate. The critical failure volume is subtracted from the current total wear volume to obtain the residual wear volume. The ratio of the residual wear volume to the current wear rate is the predicted remaining processing time.

5. The control system for improving the numerical control machining accuracy of bionic UAV wings according to claim 1 is characterized in that: The multi-body dynamics model of the five-axis machine tool is established by: A multi-body dynamics model of a five-axis machine tool is established in Simulink. The matrix forms of mass M, damping C, and stiffness K are obtained by frequency response function fitting. The spindle-guideway-screw coupled vibration equation is shown as follows: Among them, M is the mass matrix, which represents the inertia of each moving part, C is the damping matrix, which represents the energy dissipation of the system, K is the stiffness matrix, which represents the elastic restoring force of the system, and q is the displacement of each degree of freedom of the machine tool, which is expressed as x, y, z represent the linear displacement of the tool in the Cartesian coordinate system, θ A is the angular displacement around axis A, θ C is the angular displacement around the C axis, is the velocity vector containing all degrees of freedom of the machine tool, expressed as Where v represents velocity, ω represents angular velocity, is the acceleration vector containing all degrees of freedom of the machine tool, expressed as Where a represents acceleration, α represents angular acceleration, and FC represents cutting force vector.

6. The control system for improving the numerical control machining accuracy of bionic UAV wings according to claim 1, characterized in that: Dynamically divide sub-routes, the specific method is: In the machining path planning stage, the curvature gradient of each path point is calculated in real time through the second-order derivative calculation of the point cloud. The maximum allowable curvature change and the allowable material removal rate are determined through machining tests. The feed step length and feed speed are determined according to the initial setting of the process parameters of the CNC machining machine tool. The product of the feed step length and feed speed is used to express the instantaneous material removal rate. The formula is used represents the sub-route division rule, where maxk represents the maximum allowable curvature change, Δk represents the curvature change rate, and Q p Indicates the allowable material removal rate, a p Indicates the feed step length, v p Indicates feed speed, L i Indicates the length of the i-th sub-route, that is, L is equal to and The minimum value in the original continuous path is divided into L i Interval subroutes.

7. A control system for improving the numerical control machining accuracy of bionic UAV wings according to claim 6, characterized in that: Establish an error prediction model. The specific method is as follows: After the sub-route is determined, the monitoring frequency is adjusted according to the error prediction model, using the formula Represents the error prediction model, where R i represents the curvature radius of the sub-route calculated by the path geometry, k s represents the system equivalent stiffness determined by the hammering method, ΔT represents the processing temperature rise measured in real time by the infrared thermal imager, and D f Represents the acoustic emission energy, which is obtained by wavelet energy analysis of the acoustic emission signal.

8. The control system for improving the numerical control machining accuracy of bionic UAV wings according to claim 7, characterized in that: The monitoring frequency is regulated according to the error prediction model. The specific method is as follows: Frequency control function Control monitoring frequency, where f b Indicates the reference frequency, that is, the minimum effective sampling requirement of the sensor, f m represents the maximum frequency, i.e. the sensor hardware limit, τ is the accuracy standard, and γ is the precision factor.

9. A control system for improving the numerical control machining accuracy of bionic UAV wings according to claim 8, characterized in that: Correct the machining path in real time. The specific method is as follows: Set the geometric error threshold to 0.5τ. When it is greater than or equal to 0.5τ, use the formula P E =P N +ΔP performs reverse correction on the tool path, where P N Represents the coordinates of the next tool path point originally generated, ΔP represents the path correction vector, and the x-direction component is equal to The y-direction component is equal to The z-direction component is equal to where R x , R y , R z Directional components of the radius of curvature on the x-axis, y-axis, and z-axis; when When it is less than 0.5τ, it is considered normal and no processing is performed; When the monitored temperature is greater than or equal to the temperature safety threshold, the cooling system is triggered to be enhanced, using the formula Dynamically adjust the coolant flow rate according to the temperature rise, where LE b represents the standard coolant flow rate, ε is the temperature change factor, T is the monitored instantaneous temperature, maxT is the temperature safety threshold, and ΔT is the instantaneous processing temperature rise at the monitoring time point, that is, the temperature difference between the current monitoring time point and the previous monitoring time point; When the monitored temperature is lower than the temperature safety threshold, it is considered normal and no action is taken; When the acoustic emission energy D f When the fiber damage threshold D is greater than or equal to the fiber damage threshold, the next sub-route is judged to be in the damage area and the sub-route in the damage area is automatically bypassed. f When it is less than the fiber damage threshold D, the next sub-route is judged to be in the normal area and no processing is performed.

Images