Research On Graphical Parametric Programming And Machining Planning Based On Bending Robot

With the rapid development of automation in sheet metal bending,there is an increasing demand in the sheet metal industry for small-batch,multi-variety,and highly efficient flexible production.Due to the limitations of traditional manual bending in terms of scalability,robustness,and safety,research on graphic parameterization programming and machining planning based on bending robots is an important direction for achieving automation in sheet metal bending.Based on the analysis of current domestic and foreign sheet metal bending processes and system software design,this paper designs a programming system that uses graphic parameterization programming as the interaction method,with neural networks and path planning algorithms as computational support.The main focus of the research and work includes the following aspects:(1)A bending robot graphic parameterization programming system has been designed and developed.The system adopts a modular design approach,with process parameters connecting each module.It is developed using Solidworks for secondary development,Any CAD for 3D visualization engine,and Visual Studio as the development platform for the robot sheet metal bending graphic parameterization programming system.The system consists of several modules,including sheet metal part reading module,3D display module,parameter extraction module,2D simulation module,3D simulation module,and database management module.These modules are designed and developed at a fundamental level to be applied in the graphic parameterization programming system for sheet metal bending.(2)A sliding block pressing amount prediction model was constructed using a genetic algorithm optimization-based backpropagation neural network(GA-BPNN).A mathematical model of the bending process environment was established,and a sliding block pressing amount prediction model based on the GA-BP neural network algorithm was designed.The inputs of the model include the elastic modulus of the sheet,yield strength,hardening coefficient,hardening exponent,sheet thickness,lower die groove width,lower die chamfer,upper die corner radius,and bending forming angle.The output is the sliding block pressing amount.To address the issue of significant errors caused by the coupling of multiple factors,a 5-input sliding block pressing amount prediction model based on material classification was designed.This model distinguishes between material factors and process factors and optimizes the prediction model for different material scenarios.(3)A bending process feeding path planning scheme was constructed using an improved artificial potential field method based on discrete main and auxiliary computation.The positional relationship between the sheet metal part and the machine tool die during sheet metal bending was analyzed.Existing path planning algorithms were examined,and the artificial potential field method,suitable for bending process scenarios,was chosen.An improved artificial potential field method based on discrete main and auxiliary computation(DMA-APF)was proposed.It employed a discrete main and auxiliary point algorithm to match the posture of the sheet metal part.The DMA-APF method incorporated the influence of target and object distances,along with a local minimum point escape strategy based on random step lengths.Additionally,a mathematical model for bending adaptive operations was established.(4)Experimental designs were developed to address the path planning problems in both the bending graphic parameterization programming system and the bending feeding and retrieval tasks.The experimental results demonstrate the following findings: The bending graphic parameterization programming system allows for a convenient design process from the 3D model of the sheet metal part to the final processing and forming through its graphical interface.The improved artificial potential field method based on discrete main and auxiliary computation was applied to plan the feeding and retrieval of four sheet metal parts in the experiment.With a minimum safety distance of2 mm,the planning duration was 14.08 seconds.This approach effectively enhanced the safety and efficiency of the bending process.The bending process prediction model based on GA-BPNN(Genetic Algorithm-Backpropagation Neural Network)was combined with the bending follow-up trajectory model.This combination ensured that the sheet metal was safely and collision-free removed from the bending environment after the bending process.The bending step edge error was within 0.3mm,and the angle deviation of the bending forming was within 0.5°,meeting the requirements for bending process accuracy.