Search

CN-121979068-A - Large-model-driven numerical control intelligent agent executing device

CN121979068ACN 121979068 ACN121979068 ACN 121979068ACN-121979068-A

Abstract

The invention discloses a large-model-driven numerical control agent executing device, which belongs to the technical field of intelligent manufacturing and numerical control, and comprises an agent subsystem, a real-time actuator subsystem, a virtual machine tool subsystem and a physical machine tool, wherein the agent subsystem generates a verified instruction sequence through simulation interaction with the virtual machine tool subsystem, the real-time actuator subsystem is responsible for analyzing and executing an agent generated instruction to drive the physical machine tool to move, the virtual machine tool subsystem is used for verifying and optimizing the instruction of the agent subsystem to provide simulation verification support, and the physical machine tool is driven by the real-time actuator subsystem to execute final machining operation. According to the large-model-driven numerical control intelligent agent executing device, intelligent optimization of a machine tool machining process is realized through driving of the cutting force model, the control model and the error model in the intelligent agent, and machining efficiency and machining quality are improved under the condition that machining dynamic accuracy is ensured.

Inventors

  • SHEN BIN
  • HUANG YUNYING
  • CHEN SULIN

Assignees

  • 上海交通大学

Dates

Publication Date
20260505
Application Date
20260204

Claims (8)

  1. 1. The large-model driven numerical control intelligent agent executing device is characterized by comprising an intelligent agent subsystem, a real-time executor subsystem, a virtual machine tool subsystem and a physical machine tool; The intelligent body subsystem generates a verified instruction sequence through simulation interaction with the virtual machine tool subsystem, and comprises a path generator, a cutting force model function module, a control model function module, a path planner and an error compensator, wherein the path generator converts process planning or part geometric information generated by a large model layer into an initial tool path based on reinforcement learning; The path generator extracts a machining area, a layer depth, a feeding mode and a cutter type, and automatically generates rough machining, semi-finishing and finishing paths according to material properties, cutter geometric parameters and a machining strategy, and the path planning of multi-axis motion is as follows: ; Wherein, the 1 To the generated path point The axis number of the multi-axis linkage is represented, And Representing the 1 st axis and the 1 st The function of the motion of the shaft, In order to achieve a feed rate, For the radius of the tool, Is the length of the cutter; The real-time executor subsystem is responsible for analyzing and executing the instruction generated by the intelligent agent and driving the physical machine tool to move, and comprises an instruction sequence executor, a motion control module, a PLC and I/O control module, a low-level interpolator and a feedback module, wherein the instruction sequence executor uses a circulating buffer to receive and buffer the instruction sequence of the intelligent agent, analyzes the instruction sequence generated by the intelligent agent and converts the instruction sequence into an executable motion control command, is responsible for real-time execution of the command, controls the action of the machine tool and feeds key information back to the intelligent agent; The virtual machine tool subsystem is used for verifying and optimizing instructions of the intelligent agent subsystem and providing simulation verification support; the physical machine tool is driven by the real-time actuator subsystem to perform a final machining operation.
  2. 2. The large model driven numerical control intelligent agent executing device of claim 1, wherein the cutting force model function module is used for predicting instantaneous cutting force, torque and thermal influence distribution in the processing process so as to evaluate the rationality of the path and the feeding parameters; the control model functional module is used for optimizing iteration of the intelligent body, fitting control characteristics of the actual machine tool through a neural network algorithm, simulating the output of the intelligent body, and further obtaining evaluation data; the track planner is used for converting discrete path points output by the path generator into continuous executable track instructions on the premise of ensuring smoothness, processing precision and real-time performance; The error compensator corrects the geometric and dynamic errors generated in the machining process on line based on the prediction results of the cutting force model functional module and the control model functional module.
  3. 3. The large model driven numerical control intelligent agent executing device of claim 2, wherein the motion control module comprises multi-axis control, interpolation algorithm and real-time control, and realizes accurate control of motion trail; the PLC and the I/O control module realize real-time logic processing of the machine tool, ensure safe logic execution and provide a standardized instruction sequence interface; The low-level interpolator carries out fine interpolation on the input coarse interpolation track data, thereby improving the control precision; the feedback module is switched to a traditional motion control mode according to abnormal information of the feedback agent drive, and machine tool state adjustment and recovery are achieved.
  4. 4. The large model driven numerical control intelligent agent executing device of claim 3, wherein the virtual machine tool subsystem is composed of a machine tool model, a control model and a cutting force model, virtual modeling is carried out on the machine tool by collecting the motion state data of the actual machine tool, and the model state of the actual machine tool is fitted through a neural network.
  5. 5. The large model driven numerical control intelligent agent executing device of claim 4, wherein the cutting force model function module calculates cutting force, feeding force and normal component force based on the contact geometry and material characteristics of a cutter and a workpiece, and simulates the load change trend under different spindle rotation speeds and feeding rates, and the specific calculation formula is as follows: ; Wherein, the In order for the cutting force to be high, Is a unit cutting force correction value, In order to achieve a depth of cut, In order to achieve a cutting speed, the cutting speed, For the radius of the tool, The rotational speed of the main shaft is the rotational speed of the main shaft, 、 、 An index determined for the experiment.
  6. 6. The large model driven numerical control intelligent agent executing device of claim 5, wherein the control model function module simulates the motion process of the physical machine tool by calculating the periodic target current, and the calculation process of the periodic target current is as follows: ; Wherein, the For the maximum speed of the speed of rotation, At the time of maximum acceleration, the maximum acceleration is, As a result of the location of the object, Is a coefficient of proportionality and is used for the control of the power supply, As an integral coefficient of the power supply, As a result of the differential coefficient, In order to be able to carry out a cycle, Is a periodic target time.
  7. 7. The large model driven numerical control intelligent agent executing device of claim 6, wherein the track planner calculates the periodic target position to generate a smooth track meeting the constraint of acceleration and jerk, and cooperates with the control model function module to realize track sectional dynamic optimization, and the periodic target position of the multi-axis motion is as follows: ; Wherein, the At the time of maximum jerk, Is the error value caused by the cutting force.
  8. 8. The large model driven numerical control intelligent agent executing device of claim 7, wherein the error compensator calculates position errors in the machining process, models and compensates track errors caused by control and cutting force of each shaft, outputs dynamic compensation vectors to correct the track planner or control model output, and the calculation formula of the position errors is as follows: ; Wherein, the As a coefficient of speed (f) the speed, Is used as a mass coefficient of the composite material, Is the speed deviation.

Description

Large-model-driven numerical control intelligent agent executing device Technical Field The invention relates to the technical field of intelligent manufacturing and numerical control, in particular to a large-model-driven numerical control intelligent body executing device. Background The existing large-model driven numerical control system is based on traditional instruction modes such as G codes and the like, depends on fixed CAM, decoding, track interpolation and servo loops, has limited rigidity and compensation modes of functional chains, is difficult to adapt to complex process environments and dynamic disturbance, and cannot meet related interaction requirements of modern large-model intelligent agents in input and output design. The existing large model driven numerical control system has the following pain points: contradiction between real-time requirements of the numerical control device and non-real-time output of the large model; how to influence and adjust the real-time system control based on non-real-time data; How to combine the intelligent body with real-time control, so as to effectively improve the precision, quality and efficiency of the numerical control device; how to replace the manual optimization process based on agent application. Disclosure of Invention The invention aims to provide a large-model-driven numerical control intelligent agent executing device, which realizes intelligent optimization of a machine tool machining process through driving a cutting force model, a control model and an error model in an intelligent agent, and improves machining efficiency and machining quality under the condition of ensuring machining dynamic accuracy. In order to achieve the above purpose, the invention provides a large model driven numerical control intelligent agent executing device, which comprises an intelligent agent subsystem, a real-time executor subsystem, a virtual machine tool subsystem and a physical machine tool; The intelligent body subsystem generates a verified instruction sequence through simulation interaction with the virtual machine tool subsystem, and comprises a path generator, a cutting force model function module, a control model function module, a path planner and an error compensator, wherein the path generator converts process planning or part geometric information generated by a large model layer into an initial tool path based on reinforcement learning; The path generator extracts a machining area, a layer depth, a feeding mode and a cutter type, and automatically generates rough machining, semi-finishing and finishing paths according to material properties, cutter geometric parameters and a machining strategy, and the path planning of multi-axis motion is as follows: ; Wherein, the 1 To the generated path pointThe axis number of the multi-axis linkage is represented,AndRepresenting the 1 st axis and the 1 stThe function of the motion of the shaft,In order to achieve a feed rate,For the radius of the tool,Is the length of the cutter; The real-time executor subsystem is responsible for analyzing and executing the instruction generated by the intelligent agent and driving the physical machine tool to move, and comprises an instruction sequence executor, a motion control module, a PLC and I/O control module, a low-level interpolator and a feedback module, wherein the instruction sequence executor uses a circulating buffer to receive and buffer the instruction sequence of the intelligent agent, analyzes the instruction sequence generated by the intelligent agent and converts the instruction sequence into an executable motion control command, is responsible for real-time execution of the command, controls the action of the machine tool and feeds key information back to the intelligent agent; The virtual machine tool subsystem is used for verifying and optimizing instructions of the intelligent agent subsystem and providing simulation verification support; the physical machine tool is driven by the real-time actuator subsystem to perform a final machining operation. Preferably, the cutting force model function module is used for predicting instantaneous cutting force, torque and thermal influence distribution in the machining process so as to evaluate the rationality of the path and the feeding parameters; the control model functional module is used for optimizing iteration of the intelligent body, fitting control characteristics of the actual machine tool through a neural network algorithm, simulating the output of the intelligent body, and further obtaining evaluation data; the track planner is used for converting discrete path points output by the path generator into continuous executable track instructions on the premise of ensuring smoothness, processing precision and real-time performance; The error compensator corrects the geometric and dynamic errors generated in the machining process on line based on the prediction results of the cutting force model functio