New Process Development For Warm Forging Of Conjunction Gear Based On Finite Element Simulation And Neural Network Technology And Design And Optimization Of Pre-forging Part

The vigorous development of the automobile industry has continuously increased the demand for automobiles.At the same time,it has caused the improvement of the overall performance requirements of the automobile.The conjunction gear is an important gear used in vehicle gearboxes.Its quality affects the automobile engine and thus the overall performance and quality of the vehicle.Therefore,the past production methods of conjunction gear cannot meet today's needs,and a new production and processing method needs to be developed.After comprehensive consideration,it is a feasible solution to design and develop the conjunction gear of processing methods by using the warm-cold composite precision forging technology.During the design process,it was discovered that it's necessary to design a pre-forging step because the conjunction gear is a complex part and in order to improve the service life of the warm final forging die.However,there is no more unified idea of pre-forging design.Most of them are dependent on experience and constant trial and error to gradually find a more suitable pre-forging shape and size,making the design cycle quite long.Although with the development of finite element technology,finite element simulation software can be used to simulate and analyze the forging process,thereby saving a lot of trial-and-error costs.However,due to too many influencing factors in the forging process,it is difficult to find an optimal solution.Therefore,the BP neural network technology is used to model the shape and size parameters of the pre-forged parts,and the orthogonal experiment method is used to greatly reduce the number of required experiments.It can obtain appropriate experimental data quickly.First of all,by analyzing the part drawing as well as combined with the actual production conditions of the factory,the processing plan of the warm-cold composite precision forging is formulated.Then,the final forging of conjunction gear is preliminarily designed.And combined with finite element simulation software,the pre-forging is analyzed and designed based on the final forging.During this process,it was found that the different extrusion methods and whether the pre-forged part has a beveled edge would have a relatively large impact on the forming.Therefore,four schemes were designed to explore its influence on the final forging process and to determine the shape of the pre-forged part.After exploration,it is determined that the shape of the pre-forged part with the beveled edge and forward extrusion method is the most suitable shape and structure.Secondly,after determining the shape and structure of the pre-forged part,the current theoretical exploration was further analyzed based on the long-term practical experience and actual production conditions in factory.And then,some of the theoretically feasible but prone to problems in actual production were optimized and modified and the bar billet was designed.At the same time,the finite element simulation software was used to simulate and analyze the whole warm forging process,which is the process from upsetting to final forging.While verifying the feasibility of the currently designed final forging and pre-forging,it is found that if the upsetting step is omitted and the pre-forging is performed directly,the pre-forging and final forging results are almost the same as the results without upsetting step.But there is a big difference between the simulation results of the pre-forging step omitted and the final forging directly.Therefore,it is determined that the upsetting step can be omitted in the finite element simulation without affecting the results,but the pre-forging step cannot be omitted to directly perform final forging.It speeds up the process for subsequent acquisition of neural network data,while ensuring the accuracy of the experiment.Finally,the BP neural network is used for modeling to construct the nonlinear relationship between the dimensional parameters of the pre-forging part and the maximum wear depth of the final forging die.In order to further reduce the duration of the simulation experiment,so orthogonal experiments are used to obtain correct and effective data.After training,it is verified that the neural network has a reliable prediction effect,and its average relative error is only 2.49%.And then,the optimal pre-forging shape and size parameters are found by using it.After verification by finite element simulation,the credibility of the results is proved.This design idea reduces a large number of experiments,effectively improves the efficiency of process development,and provides a reliable basis for the process development process of forging.