A boring and milling machine thermal error compensation system, method, terminal and medium

Abstract

The present application belongs to the field of gantry boring and milling machine processing technology, and specifically discloses a boring and milling machine thermal error compensation system, method, terminal and medium, which comprises: synchronously collecting the power of the main motor of the machine tool, the pressure of the hydraulic oil cylinder, the spatial coordinates of the ram and the Z-direction extension length, and performing space-time alignment; converting the spatial coordinate features into a structure stiffness penalty vector reflecting the deflection deformation of the beam and the overturning moment of the ram, inputting the structure stiffness penalty vector into a semantic segmentation network with a spatial prior attention mechanism to identify the processing state sequence; constructing a white-box heat source model decomposed into cutting work, hydraulic work and time-varying spatial friction work, and deducing the dynamic heat source intensity distribution; calculating the real-time thermal error amount based on the thermal elastic deformation mechanism and converting it into a coordinate correction bias, and actively optimizing and feedback controlling under the constraint of physical deformation boundary combined with process parameters. The present application realizes the collaborative thermal error compensation of source heat inhibition and dynamic coordinate hedging under the safety boundary of stiffness.

Description

Technical Field

[0001] This invention belongs to the field of gantry milling machine machining technology, specifically relating to a thermal error compensation system, method, terminal, and medium for milling machines. Background Technology

[0002] As the mother machines of industry, the machining accuracy of machine tools directly determines the manufacturing level of high-end equipment. Especially in the machining of large and long parts such as wind turbine spindles, large CNC fixed beam gantry milling machines generate a lot of heat during long-term continuous operation due to their long machining cycle and complex power system.

[0003] Existing CNC gantry milling machines, when machining large workpieces, are generally equipped with high-power spindles, long-stroke lead screw drives, and dual hydraulic cylinders to balance the weight of the ram. To reduce thermal deformation errors during machining, current conventional thermal error compensation methods mainly involve placing multiple temperature sensors on key structural components of the machine tool to collect surface temperature data in real time. Based on this, the system calculates the theoretical thermal deformation under the current temperature distribution using a model, and then sends this as a compensation offset command to the CNC system to perform reverse correction of the machining coordinates to guide the actual operation.

[0004] However, existing thermal error compensation techniques face the following limitations in practical applications: First, conventional methods rely excessively on external temperature sensors, resulting in an inherent "thermal inertia hysteresis" phenomenon. Sensors can only measure the temperature already conducted to the structural surface and cannot synchronously sense the heat source at the moment of its generation. Furthermore, during long-stroke machining on large fixed-beam gantry milling machines, changes in the ram overhang length and the beam span cause nonlinear, time-varying frictional work, leading to drastic dynamic changes in the internal heat source distribution with spatial orientation. Fixed sensor measurement points struggle to capture such complex spatiotemporal physical changes, resulting in significant time lag and spatial deviation in thermal error prediction.

[0005] Secondly, most existing thermal error models are black-box models based on offline data fitting, which sever the underlying physical connection between machine tool energy consumption and heat generation. This approach fails to transform the actual cutting work, long-stroke nonlinear friction work, and hydraulic system auxiliary work of the machine tool into explicit dynamic internal heat sources, ultimately limiting further improvements in the accuracy and response time of the overall machine thermal error compensation. Summary of the Invention

[0006] This invention addresses the problems in the prior art by providing a thermal error compensation system, method, terminal, and medium for boring and milling machines. It solves the problem that conventional methods in the background rely excessively on external temperature sensors, resulting in inherent "thermal inertia hysteresis" and causing significant time lag and spatial deviation in thermal error prediction. Simultaneously, it addresses the issue that most existing thermal error models are black-box models based on offline data fitting, severing the underlying physical connection between machine tool energy consumption and heat generation, thus limiting further improvements in the accuracy and response time of overall machine thermal error compensation.

[0007] The technical solution adopted in this invention is as follows: In a first aspect, this application provides a method for compensating for thermal errors in boring and milling machines, comprising the following steps: Step S1: During the workpiece machining process, the real-time timing power signal of the machine tool's main motor is synchronously acquired through the multi-source status monitoring unit. Dynamic pressure signal of ram hydraulic cylinder And the spatial coordinates of the ram on the worktable plane and the crossbeam. The extension length of the ram in the Z direction ; Step S2: Based on the time-varying stiffness characteristics of the machine tool structure, the time-series power signal W(t) and the dynamic pressure signal are... Spatial coordinates and the length of the extension Perform spatiotemporal feature alignment to construct a joint feature sequence of physical stiffness constraints; Step S3: Input the physical stiffness constraint joint feature sequence into a semantic segmentation network with a spatial prior attention mechanism to identify and output the current actual machining operation state sequence of the machine tool and its corresponding time node division; Step S4: Construct a heat source calculation model for nonlinear friction and hydraulic coupling of boring and milling machine. Based on the actual machining operation state sequence, calculate the machine tool prediction power in each state time period. Calculate the cutting heat source intensity and time-varying friction heat source intensity based on the power-heat conversion mechanism to obtain the dynamic heat source intensity distribution of boring and milling machine in the current spatial pose. Step S5: Based on the time node division, according to the heat accumulation and dissipation state of the dynamic heat source intensity distribution in the actual processing operation state sequence, calculate the real-time thermal error in the machining axis direction of the boring and milling machine based on the thermoelastic deformation mechanism, convert the real-time thermal error into coordinate correction offset, and feed it back to the feed loop of the boring and milling machine CNC system in real time for dynamic coordinate offset compensation.

[0008] Furthermore, the specific process of constructing the joint feature sequence of physical stiffness constraints in step S2 is as follows: The dynamic pressure signal is obtained by linear interpolation. Spatial coordinates and the length of the extension Upsampled to the time-series power signal Same frequency; For each time step The spatial coordinate characteristics are transformed into a structural stiffness penalty vector that reflects the deflection deformation of the beam and the overturning moment of the ram. ,in The coordinates of the physical center point of the beam guide rail; The upsampled time-series power signal The dynamic pressure signal With the structural stiffness penalty vector The sequences are spliced ​​together to generate the joint feature sequence of the physical stiffness constraints.

[0009] Furthermore, in the heat source calculation model described in step S4, the machine tool predicted power... Including cutting work items Hydraulic maintenance work items and the work done by time-varying spatial friction ,Right now ; The rules are as follows: The cutting work item ,in The specific cutting force parameter of the workpiece material. , , These are the real-time depth of cut, feed rate, and cutting speed, respectively. The hydraulic maintenance work item ,in For real-time dynamic pressure of hydraulic cylinders, The rated displacement of the hydraulic pump. This refers to the volumetric efficiency parameter of the hydraulic system. The work done by time-varying spatial friction ,in The initial friction coefficient of the guide rail is . The deflection friction gain coefficient caused by deviation from the center of the beam. The gain coefficient of overturning friction caused by the extension of the ram. For the driving force of the feed system, For feed rate; The formula , , , , All are pre-calibrated known physical constants; When calculating the dynamic heat source intensity distribution, the cutting work term is included. Work done by friction with time-varying space Multiply by the corresponding work-heat conversion coefficients to obtain the corresponding cutting heat source intensity and time-varying friction heat source intensity.

[0010] Furthermore, the semantic segmentation network with spatial prior attention mechanism described in step S3 uses the structural stiffness penalty vector described in step S2 for weight allocation. Generate a dynamic attention bias matrix; The semantic segmentation network includes a temporal convolutional encoder and a self-attention module; When calculating the self-attention weight matrix, and The spatial feature attenuation factor is mapped and introduced as a penalty bias term into the temporal attention score matrix; When the coordinates of the sliding ram Deviation The distance increases, or the Z-axis extension length increases. As the pressure increases, the network reduces the weight of high-frequency power fluctuation features in classification and increases the weight of pressure signals. Weights in state determination.

[0011] Furthermore, the semantic segmentation network employs an adaptive energy conservation loss function during training. :

[0012] in, The total number of sampling points. This represents the total number of actual processing and operation status segments output by the network. For the first The start and end times of each state segment. This represents the actual power collected. For the machine tool in the first The actual integral electrical energy consumed within each state segment To prevent the use of tiny constants with a denominator of zero; the loss function is obtained through... The energy error term is transformed into a relative deviation rate, enabling the energy constraints of different processing states to achieve dimensionless adaptive balance.

[0013] Furthermore, the calculation of the real-time thermal error in step S5 is forced to include active optimization of process parameters under the physical deformation boundary constraints of the fixed-beam gantry structure. These boundary constraints specifically include: Ram overturning moment constraint: ,in The optimized balanced pressure for the hydraulic cylinder The area of ​​the hydraulic cylinder under stress. and These are the lever arm lengths from the point of force application on the balance cylinder and the point of force application on the cutting cylinder to the center of mass of the slide block, respectively. To the current extension length The maximum cutting overturning moment; While calculating the thermal error, the optimization objectives are to minimize the energy consumption for material removal per unit volume and the amount of ineffective heat source generation. Under the premise of satisfying the physical deformation boundary constraints, a multivariate parameter optimization algorithm is used to iteratively output the optimal cutting speed. Feed rate and with the present Matching hydraulic balance compensation pressure And feed it back to the CNC system.

[0014] Secondly, this application provides a boring and milling machine thermal error compensation system to implement the boring and milling machine thermal error compensation method described in the first aspect, comprising: The boring and milling machine's execution components include a bed, a worktable, a column, a crossbeam, a ram, and an X-axis transmission mechanism for driving the worktable to move on the bed, a Y-axis transmission mechanism for driving the ram to move on the crossbeam, and a Z-axis transmission mechanism for driving the ram to move in the Z-direction; a hydraulic cylinder for balancing its own weight is connected to the ram. The multi-source status monitoring unit is connected to the boring and milling machine's execution component and is configured to synchronously acquire the real-time timing power signal of the machine tool's main motor during workpiece machining. The real-time dynamic pressure signal of the hydraulic cylinder and the spatial coordinates of the slide block. The extension length of the slide in the Z direction ; The state recognition and prediction processor is communicatively connected to the multi-source state monitoring unit and transmits the time-series power signal. Dynamic pressure signal Spatial coordinates and the length of the extension Spatiotemporal feature alignment is performed and a joint feature sequence of physical stiffness constraints is constructed. The operation state sequence is identified and processed based on a semantic segmentation network with a spatial prior attention mechanism. The dynamic heat source intensity distribution is obtained through a heat source calculation model. The thermal error compensation controller is communicatively connected to the state recognition and prediction processor and the boring and milling machine CNC system. It is configured to calculate the real-time thermal error based on the dynamic heat source intensity distribution and state time nodes, convert it into coordinate correction offset, and feed it back to the feed loop of the CNC system in real time for dynamic coordinate offset compensation.

[0015] Furthermore, the X-axis transmission mechanism, Y-axis transmission mechanism, and Z-axis transmission mechanism are all lead screw transmission pairs with a double nut preload structure.

[0016] Thirdly, this application provides a terminal, including: The memory is used to store the thermal error compensation program for the boring and milling machine; A processor is configured to implement the steps of the boring and milling machine thermal error compensation method as described in the first aspect when executing the boring and milling machine thermal error compensation program.

[0017] Fourthly, this application provides a computer-readable storage medium that stores computer instructions. When a computer reads the computer instructions in the storage medium, the computer executes the boring and milling machine thermal error compensation method as described in the first aspect.

[0018] As can be seen from the above technical solutions, the advantages of the present invention are: By synchronously collecting time-series power, dynamic pressure, multi-axis spatial coordinates, and ram extension length, and based on spatiotemporal feature alignment and operational state segmentation, a heat source calculation model is constructed to derive the dynamic heat source intensity distribution. Then, based on the thermoelastic deformation mechanism, the real-time thermal error is calculated and converted into coordinate correction bias, realizing dynamic coordinate offset thermal error compensation from energy consumption, friction mechanism bottom layer to CNC feed circuit.

[0019] By employing linear interpolation to perform high-frequency upsampling and alignment of multi-source heterogeneous signals, and transforming real-time spatial coordinate features into structural stiffness penalty vectors that reflect beam deflection and ram overturning moment, the underlying decoupling and fusion of physical deformation mechanism and data-driven features are achieved. This provides physical stiffness constraint inputs with pose characteristics for accurate division of subsequent processing states.

[0020] By decomposing the predicted power of the machine tool into cutting work, hydraulic maintenance work and time-varying spatial friction work, and explicitly introducing the deflection and overturning friction gain coefficients determined by spatial coordinates to derive the heat source intensity distribution, a digital quantitative characterization of the spatiotemporal generation characteristics of local internal heat sources is realized. This can accurately reflect the dynamic evolution law of physical heat sources caused by nonlinear friction in long-stroke machining scenarios of fixed beam gantry milling machines.

[0021] By using the structural stiffness penalty vector to generate a dynamic attention bias matrix, a spatial feature attenuation factor is mapped when the ram deviates from the center of the crossbeam guide rail or when the Z-direction overhang becomes longer. This factor is then introduced as a penalty bias term into the time attention score matrix to intervene in the self-attention weight allocation. This achieves physical layer compensation for the stress deformation and chatter interference of large heavy-duty boring and milling machines, which can suppress signal distortion and eliminate misclassification of state caused by mechanical deformation burrs.

[0022] By employing an adaptive energy conservation loss function that combines the prediction error of the connection point with the total energy consumption deviation of the state segment in the semantic segmentation network, and by using real integrated electrical energy to transform the energy error term into a dimensionless relative deviation rate, an adaptive balance of energy conservation constraints for different processing dynamic elements is achieved. This avoids the black-box hyperparameter interference caused by manually setting constant weights during network training.

[0023] By introducing the physical boundary constraint of the ram overturning moment unique to the fixed beam gantry structure, and by iteratively optimizing with the goal of minimizing energy consumption for material removal per unit volume and minimizing the amount of ineffective heat source generation, the coordinated active control of process parameters and hydraulic balance pressure is achieved under the premise of meeting the rigidity safety and mechanical life of heavy machine tools. This can suppress the temperature rise and thermal error accumulation of machine tools caused by ineffective work from the source.

[0024] By configuring specific mechanical actuators, multi-source state monitoring units, state recognition and prediction processors, and thermal error compensation controllers within the system, a closed-loop linkage of hardware and software is achieved for mechanical components, heterogeneous sensors, state segmentation algorithms, and coordinate offset actuators. This enhances the system-level thermal stability and comprehensive thermal error defense capabilities of heavy-duty fixed-beam gantry milling machines during long-term continuous machining operations.

[0025] By limiting the X-axis, Y-axis, and Z-axis transmission mechanisms to lead screw transmission pairs with double nut preload structures, the internal clearance of the transmission system is effectively eliminated and the mechanical contact rigidity is significantly improved. This reduces reverse transmission friction fluctuations and suppresses frictional heat generation, thereby further improving the motion response capability and reverse positioning stability of the entire machine during dynamic coordinate compensation execution. Attached Figure Description

[0026] To more clearly illustrate the technical solution of the present invention, the accompanying drawings used in the description will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0027] Figure 1 This is a flowchart of a boring and milling machine thermal error compensation method according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the thermal error compensation system for a boring and milling machine according to an embodiment of the present invention.

[0028] In the diagram: 1. Bed; 2. X-axis transmission mechanism; 3. Worktable; 4. Column; 5. Crossbeam; 6. Z-axis transmission mechanism; 7. Hydraulic cylinder; 8. Y-axis transmission mechanism. Detailed Implementation

[0029] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0030] Please see Figure 1 As shown, this application provides a method for thermal error compensation of a boring and milling machine, which includes the following steps: Step S1: During the workpiece machining process, the real-time timing power signal of the machine tool's main motor is synchronously acquired through the multi-source status monitoring unit. Dynamic pressure signal of ram hydraulic cylinder And the spatial coordinates of the ram on the worktable plane and the crossbeam. The extension length of the ram in the Z direction ; Step S2: Based on the time-varying stiffness characteristics of the machine tool structure, the time-series power signal W(t) and the dynamic pressure signal are... Spatial coordinates and the length of the extension Perform spatiotemporal feature alignment to construct a joint feature sequence of physical stiffness constraints; Step S3: Input the physical stiffness constraint joint feature sequence into a semantic segmentation network with a spatial prior attention mechanism to identify and output the current actual machining operation state sequence of the machine tool and its corresponding time node division; Step S4: Construct a heat source calculation model for nonlinear friction and hydraulic coupling of boring and milling machine. Based on the actual machining operation state sequence, calculate the machine tool prediction power in each state time period. Calculate the cutting heat source intensity and time-varying friction heat source intensity based on the power-heat conversion mechanism to obtain the dynamic heat source intensity distribution of boring and milling machine in the current spatial pose. Step S5: Based on the time node division, according to the heat accumulation and dissipation state of the dynamic heat source intensity distribution in the actual processing operation state sequence, calculate the real-time thermal error in the machining axis direction of the boring and milling machine based on the thermoelastic deformation mechanism, convert the real-time thermal error into coordinate correction offset, and feed it back to the feed loop of the boring and milling machine CNC system in real time for dynamic coordinate offset compensation.

[0031] In some embodiments, the specific process of constructing the joint feature sequence of physical stiffness constraints in step S2 is as follows: The dynamic pressure signal is obtained by linear interpolation. Spatial coordinates and the length of the extension Upsampled to the time-series power signal Same frequency; For each time step The spatial coordinate characteristics are transformed into a structural stiffness penalty vector that reflects the deflection deformation of the beam and the overturning moment of the ram. ,in The coordinates of the physical center point of the beam guide rail; The upsampled time-series power signal The dynamic pressure signal With the structural stiffness penalty vector The sequences are spliced ​​together to generate the joint feature sequence of the physical stiffness constraints.

[0032] In some embodiments, in the heat source calculation model described in step S4, the machine tool predicted power Including cutting work items Hydraulic maintenance work items and the work done by time-varying spatial friction ,Right now ; The rules are as follows: The cutting work item ,in The specific cutting force parameter of the workpiece material. , , These are the real-time depth of cut, feed rate, and cutting speed, respectively. The hydraulic maintenance work item ,in For real-time dynamic pressure of hydraulic cylinders, The rated displacement of the hydraulic pump. This refers to the volumetric efficiency parameter of the hydraulic system. The work done by time-varying spatial friction ,in The initial friction coefficient of the guide rail is . The deflection friction gain coefficient caused by deviation from the center of the beam. The gain coefficient of overturning friction caused by the extension of the ram. For the driving force of the feed system, For feed rate; The formula , , , , All of them are pre-calibrated known physical constants; When calculating the dynamic heat source intensity distribution, the cutting work term is included. Work done by friction with time-varying space Multiply by the corresponding work-heat conversion coefficients to obtain the corresponding cutting heat source intensity and time-varying friction heat source intensity.

[0033] In some embodiments, the semantic segmentation network with spatial prior attention mechanism described in step S3 uses the structural stiffness penalty vector described in step S2 for weight allocation. Generate a dynamic attention bias matrix; The semantic segmentation network includes a temporal convolutional encoder and a self-attention module; When calculating the self-attention weight matrix, and The spatial feature attenuation factor is mapped and introduced as a penalty bias term into the temporal attention score matrix; When the coordinates of the sliding ram Deviation The distance increases, or the Z-axis extension length increases. As the pressure increases, the network reduces the weight of high-frequency power fluctuation features in classification and increases the weight of pressure signals. Weights in state determination.

[0034] In some embodiments, the semantic segmentation network employs an adaptive energy conservation loss function during training. :

[0035] in, The total number of sampling points. This represents the total number of actual processing and operation status segments output by the network. For the first The start and end times of each state segment. This represents the actual power collected. For the machine tool in the first The actual integral electrical energy consumed within each state segment To prevent the use of tiny constants with a denominator of zero; the loss function is obtained through... The energy error term is transformed into a relative deviation rate, enabling the energy constraints of different processing states to achieve dimensionless adaptive balance.

[0036] In some embodiments, the calculation of the real-time thermal error in step S5 is forced to include active optimization of process parameters under the physical deformation boundary constraints of the fixed beam gantry structure. The specific boundary constraints include: Ram overturning moment constraint: ,in The optimized balanced pressure for the hydraulic cylinder The area of ​​the hydraulic cylinder under stress. and These are the lever arm lengths from the point of force application on the balance cylinder and the point of force application on the cutting cylinder to the center of mass of the slide block, respectively. To the current extension length The maximum cutting overturning moment; While calculating the thermal error, the optimization objectives are to minimize the energy consumption for material removal per unit volume and the amount of ineffective heat source generation. Under the premise of satisfying the physical deformation boundary constraints, a multivariate parameter optimization algorithm is used to iteratively output the optimal cutting speed. Feed rate and with the present Matching hydraulic balance compensation pressure And feed it back to the CNC system.

[0037] Please see Figure 2 As shown, this application provides a thermal error compensation system for boring and milling machines, comprising: The boring and milling machine's execution components include a bed 1, a worktable 3, a column 4, a crossbeam 5, a ram, and an X-axis transmission mechanism 2 for driving the worktable 3 to move on the bed 1, a Y-axis transmission mechanism 8 for driving the ram to move on the crossbeam 5, and a Z-axis transmission mechanism 6 for driving the ram to move in the Z direction; a hydraulic cylinder 7 for balancing its own weight is connected to the ram. The multi-source status monitoring unit is connected to the boring and milling machine's execution component and is configured to synchronously acquire the real-time timing power signal of the machine tool's main motor during workpiece machining. The real-time dynamic pressure signal of the hydraulic cylinder 7 and the spatial coordinates of the slide block. The extension length of the slide in the Z direction ; The state recognition and prediction processor is communicatively connected to the multi-source state monitoring unit and transmits the time-series power signal. Dynamic pressure signal Spatial coordinates and the length of the extension Spatiotemporal feature alignment is performed and a joint feature sequence of physical stiffness constraints is constructed. The operation state sequence is identified and processed based on a semantic segmentation network with a spatial prior attention mechanism. The dynamic heat source intensity distribution is obtained through a heat source calculation model. The thermal error compensation controller is communicatively connected to the state recognition and prediction processor and the boring and milling machine CNC system. It is configured to calculate the real-time thermal error based on the dynamic heat source intensity distribution and state time nodes, convert it into coordinate correction offset, and feed it back to the feed loop of the CNC system in real time for dynamic coordinate offset compensation.

[0038] In some embodiments, the X-axis transmission mechanism 2, the Y-axis transmission mechanism 8, and the Z-axis transmission mechanism 6 are all lead screw transmission pairs with a double nut preload structure.

[0039] In some embodiments, this application provides a terminal, including: The memory is used to store the thermal error compensation program for the boring and milling machine; A processor is configured to implement the steps of the boring and milling machine thermal error compensation method when executing the boring and milling machine thermal error compensation system.

[0040] In some embodiments, this application provides a computer-readable storage medium that stores computer instructions. When a computer reads the computer instructions in the storage medium, the computer executes the boring and milling machine thermal error compensation method.

[0041] The above description is merely a preferred embodiment of one or more embodiments of this specification and is not intended to limit the scope of one or more embodiments of this specification. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of one or more embodiments of this specification should be included within the protection scope of one or more embodiments of this specification.

Claims

1. A method for compensating thermal errors in a boring and milling machine, characterized in that, Includes the following steps: Step S1: During the workpiece machining process, the real-time timing power signal of the machine tool's main motor is synchronously acquired through the multi-source status monitoring unit. Dynamic pressure signal of ram hydraulic cylinder And the spatial coordinates of the ram on the worktable plane and the crossbeam. The extension length of the ram in the Z direction ; Step S2: Based on the time-varying stiffness characteristics of the machine tool structure, the time-series power signal W(t) and the dynamic pressure signal are... Spatial coordinates and the length of the extension Perform spatiotemporal feature alignment to construct a joint feature sequence of physical stiffness constraints; Step S3: Input the physical stiffness constraint joint feature sequence into a semantic segmentation network with a spatial prior attention mechanism to identify and output the current actual machining operation state sequence of the machine tool and its corresponding time node division; Step S4: Construct a heat source calculation model for nonlinear friction and hydraulic coupling of boring and milling machine. Based on the actual machining operation state sequence, calculate the machine tool prediction power in each state time period. Calculate the cutting heat source intensity and time-varying friction heat source intensity based on the power-heat conversion mechanism to obtain the dynamic heat source intensity distribution of boring and milling machine in the current spatial pose. Step S5: Based on the time node division, according to the heat accumulation and dissipation state of the dynamic heat source intensity distribution in the actual processing operation state sequence, calculate the real-time thermal error in the machining axis direction of the boring and milling machine based on the thermoelastic deformation mechanism, convert the real-time thermal error into coordinate correction offset, and feed it back to the feed loop of the boring and milling machine CNC system in real time for dynamic coordinate offset compensation.

2. The boring and milling machine thermal error compensation method according to claim 1, characterized in that, The specific process of constructing the joint feature sequence of physical stiffness constraints in step S2 is as follows: The dynamic pressure signal is obtained by linear interpolation. Spatial coordinates and the length of the extension Upsampled to the time-series power signal Same frequency; For each time step The spatial coordinate characteristics are transformed into a structural stiffness penalty vector that reflects the deflection deformation of the beam and the overturning moment of the ram. ,in The coordinates of the physical center point of the beam guide rail; The upsampled time-series power signal The dynamic pressure signal With the structural stiffness penalty vector The sequences are spliced ​​together to generate the joint feature sequence of the physical stiffness constraints.

3. The boring and milling machine thermal error compensation method according to claim 2, characterized in that, In the heat source calculation model described in step S4, the predicted power of the machine tool is... Including cutting work items Hydraulic maintenance work items and the work done by time-varying spatial friction ,Right now ; The rules are as follows: The cutting work item ,in The specific cutting force parameter of the workpiece material. , , These are the real-time depth of cut, feed rate, and cutting speed, respectively. The hydraulic maintenance work item ,in For real-time dynamic pressure of hydraulic cylinders, The rated displacement of the hydraulic pump. This refers to the volumetric efficiency parameter of the hydraulic system. The work done by time-varying spatial friction ,in The initial friction coefficient of the guide rail is . The deflection friction gain coefficient caused by deviation from the center of the beam. The gain coefficient of overturning friction caused by the extension of the ram. For the driving force of the feed system, For feed rate; The formula , , , , All are pre-calibrated known physical constants; When calculating the dynamic heat source intensity distribution, the cutting work term is included. Work done by friction with time-varying space Multiply by the corresponding work-heat conversion coefficients to obtain the corresponding cutting heat source intensity and time-varying friction heat source intensity.

4. The boring and milling machine thermal error compensation method according to claim 3, characterized in that, The semantic segmentation network with spatial prior attention mechanism described in step S3 uses the structural stiffness penalty vector described in step S2 for weight allocation. Generate a dynamic attention bias matrix; The semantic segmentation network includes a temporal convolutional encoder and a self-attention module; When calculating the self-attention weight matrix, and The spatial feature attenuation factor is mapped and introduced as a penalty bias term into the temporal attention score matrix; When the coordinates of the sliding ram Deviation The distance increases, or the Z-axis extension length increases. As the pressure increases, the network reduces the weight of high-frequency power fluctuation features in classification and increases the weight of pressure signals. Weights in state determination.

5. The boring and milling machine thermal error compensation method according to claim 4, characterized in that, The semantic segmentation network employs an adaptive energy conservation loss function during training. : in, The total number of sampling points. This represents the total number of actual processing and operation status segments output by the network. For the first The start and end times of each state segment. This represents the actual power collected. For the machine tool in the first The actual integral electrical energy consumed within each state segment To prevent the use of tiny constants with a denominator of zero; the loss function is obtained through... The energy error term is transformed into a relative deviation rate, enabling the energy constraints of different processing states to achieve dimensionless adaptive balance.

6. The boring and milling machine thermal error compensation method according to claim 3, characterized in that, The calculation of real-time thermal error in step S5 is forced to include active optimization of process parameters under the physical deformation boundary constraints of the fixed beam gantry structure. The specific boundary constraints include: Ram overturning moment constraint: ,in The optimized balanced pressure for the hydraulic cylinder The area of ​​the hydraulic cylinder under stress. and These are the lever arm lengths from the point of force application on the balance cylinder and the point of force application on the cutting cylinder to the center of mass of the slide block, respectively. To the current extension length The maximum cutting overturning moment; While calculating the thermal error, the optimization objectives are to minimize the energy consumption for material removal per unit volume and the amount of ineffective heat source generation. Under the premise of satisfying the physical deformation boundary constraints, a multivariate parameter optimization algorithm is used to iteratively output the optimal cutting speed. Feed rate and with the present Matching hydraulic balance compensation pressure And feed it back to the CNC system.

7. A boring and milling machine thermal error compensation system, used to implement the boring and milling machine thermal error compensation method as described in any one of claims 1-6, characterized in that, include: The boring and milling machine's execution components include a bed, a worktable, a column, a crossbeam, a ram, and an X-axis transmission mechanism for driving the worktable to move on the bed, a Y-axis transmission mechanism for driving the ram to move on the crossbeam, and a Z-axis transmission mechanism for driving the ram to move in the Z-direction; a hydraulic cylinder for balancing its own weight is connected to the ram. The multi-source status monitoring unit is connected to the boring and milling machine's execution component and is configured to synchronously acquire the real-time timing power signal of the machine tool's main motor during workpiece machining. The real-time dynamic pressure signal of the hydraulic cylinder and the spatial coordinates of the slide block. The extension length of the slide in the Z direction ; The state recognition and prediction processor is communicatively connected to the multi-source state monitoring unit and transmits the time-series power signal. Dynamic pressure signal Spatial coordinates and the length of the extension Spatiotemporal feature alignment is performed and a joint feature sequence of physical stiffness constraints is constructed. The operation state sequence is identified and processed based on a semantic segmentation network with a spatial prior attention mechanism. The dynamic heat source intensity distribution is obtained through a heat source calculation model. The thermal error compensation controller is communicatively connected to the state recognition and prediction processor and the boring and milling machine CNC system. It is configured to calculate the real-time thermal error based on the dynamic heat source intensity distribution and state time nodes, convert it into coordinate correction offset, and feed it back to the feed loop of the CNC system in real time for dynamic coordinate offset compensation.

8. The boring and milling machine thermal error compensation system according to claim 7, characterized in that, The X-axis transmission mechanism, Y-axis transmission mechanism and Z-axis transmission mechanism are all lead screw transmission pairs with a double nut preload structure.

9. A terminal, characterized in that, include: The memory is used to store the thermal error compensation program for the boring and milling machine; A processor is configured to implement the steps of the boring and milling machine thermal error compensation method as described in any one of claims 1-6 when executing the boring and milling machine thermal error compensation device.

10. A computer-readable storage medium, characterized in that, The storage medium stores computer instructions. When the computer reads the computer instructions in the storage medium, the computer executes the boring and milling machine thermal error compensation method as described in any one of claims 1-6.

Images