Research On Multi-Objective Optimization Of Aluminum Profile Extrusion Process Parameters Based On MOEA/D

With the wide application of aluminum profiles in domestic construction,automobile,manufacturing and other industries,China has become the world's la rgest aluminum profile producing area and consumer market.Extrusion is the core link in the production of aluminum profiles,and its process parameters directly determine the production cost,efficiency and energy consumption of aluminum profiles.This is closely related to the policy requirements of "Made in China 2025" and the development needs of enterprises,so the extrusion process parameters are optimized Making multiple target values optimal has become a hot spot in the field of aluminum profile research.Traditional methods rely on experience or physical formulas to establish the mapping relationship between process parameters and optimization goals,and it is very difficult to quickly optimize process parameters based on recent production conditions.With the advent of the Industrial Internet and the era of big data,aluminum profile production data is collected and stored in the enterprise database,providing necessary conditions for data-driven aluminum profile extrusion process optimization research.In this paper,aluminum profile extrusion machine is the research object,aiming at the problems of high energy consumption,low yield and serious mold wear in aluminum profile extrusion production,the research is based on extrusion production data.First,analyze the influence of extrusion process parameters on the energy consumption,yield and die wear of aluminum profile production.Secondly,a prediction model is constructed based on Gated Recurrent Unit(GRU),so as to accurately and quickly co nstruct the mapping relationship between process parameters and optimization goals.Then,based on the decomposition-based multi-objective optimization algorithm(MOEA/D),combined with the predictive model,multi-objective optimization is carried out with the goals of low energy consumption,high yield,and low die wear to solve the optimal extrusion production process parameter set.Finally,based on the enterprise energy management system,the forecasting and optimization functions are compiled into inde pendent system modules and embedded in the system to realize digital,intelligent,and automated aluminum extrusion production.The research work of this article is as follows:(1)According to the production process of aluminum profiles,first analyze the extrusion press production system and work flow,then conduct a detailed analysis of the direct and indirect factors affecting extrusion production,and finally systematically analyze the impact of extrusion process parameters on energy consumption and fi nished products.The relationship between rate and mold wear.(2)Aiming at the difficulty in modeling the physical prediction model of aluminum extrusion production and the low prediction accuracy,a neural network model to improve GRU is proposed.The model adds the Attention Model(AM)to the classic GRU network,and uses the Gravitational Search Algorithm(GSA)to improve the prediction accuracy of the model.Construct a predictive model with process parameters as input and energy consumption,yield and mold wear as output.The experimental results show that the improved model has smaller root mean square error and absolute average error than mainstream prediction models,and has good accuracy.(3)Aiming at the difficulty of multi-objective optimization of aluminum profile extrusion process parameters under multiple constraints,an improved MOEA/D algorithm is proposed for multi-objective optimization of aluminum profile extrusion process parameters.Several experiments were conducted to compare the impr oved MOEA/D with popular multi-objective optimization algorithms,and the results proved that the algorithm has better performance.At the same time,the algorithm parameters of the improved MOEA/D are tuned,so as to find the excellent algorithm parameter s to be used in the optimization of the extrusion process parameters.Finally,according to the statistical production data,the energy consumption of the optimized process parameters for aluminum extrusion production has been reduced by 13.68%,the yield rate has increased by 12.58%,and the mold wear has been reduced by 12.87%.(4)Aiming at the lack of the multi-objective optimization function of process parameters in the enterprise energy management system,the design and development of aluminum extrusion production forecasting and multi-objective optimization modules are proposed to achieve standardized interface calls and distributed cluster deployment,which improves the concurrency and availability of the system And expandability.