Intelligent Manufacturing System Performance Evaluation Modeling Based On Machine Learning

Intelligent manufacturing is increasingly becoming a major trend and core content of the future manufacturing industry.It is an important measure for enterprises to step forward to the middle and high end and accelerate the transformation and upgrading.It is also an inevitable choice for enterprises to build competitive advantages under the new normal conditions.However,many enterprises are not clear about which level they are in intelligent manufacturing and how to implement intelligent manufacturing step by step.The imbalance between supply and demand and confusion have greatly affected the effective landing of intelligent manufacturing.This paper is based on machine learning to find an intelligent evaluation method,which provides a new idea for the research of intelligent manufacturing system evaluation,and has good theoretical significance and application value.The paper's main research work and results are as follows:(1)The research focus of manufacturing system evaluation is analyzed by bibliometrics,which referrs to the guidelines for the construction of national intelligent manufacturing standard system,NIST intelligent manufacturing ecosystem and German industry 4.0reference system.The comprehensive design of the evaluation system such as dimension,index and measurement method is carried out for intelligent manufacturing,and the hierarchical analysis of noise data is carried out,and the sample data is expanded by using SeqGAN to generate countermeasure network.The construction and training of SeqGAN model are realized based on python programming.(2)The selection of neural network type,the performance index,the multi-objective genetic optimization algorithm,the cost function,the training method,the weight updating method,the coding method and so on are carried out.According to the sample matrix,the code is vectorized to improve code simplicity and improve efficiency.The modeling,training,optimization and testing of feedforward BP neural network and recursive Elman neural network optimized by NSGA-II multi-objective genetic algorithm are realized by Matlabprogramming.The orthogonal experiment and adjustment of superparameters are carried out by Minitab.(3)The two kinds of neural networks are tested and evaluated,through comparing and analyzing the aspects of network properties,topology,parameter updating methods and so on.The results show that the GABP neural network has the ability of nonlinear input-output mapping,the high robustness of generalization fault-tolerant and large-scale parallel distributed architecture,which can effectively recognize the pattern of manufacturing systems with high feature dimension.(4)A typical enterprise case study is carried out.Through the self evaluation of the enterprise,the gap analysis with the poor and excellent enterprises,and the horizontal comparative analysis of the enterprise,the overall planning of the enterprise intelligent manufacturing system is carried out.The feasibility and validity of the evaluation model are proved.The innovation of this paper is that the neural network model of intelligent manufacturing system evaluation which based on multi-objective genetic algorithm optimization is constructed with the integrated design of intelligent manufacturing evaluation system and SeqGAN generation antagonistic network.On the basis of the research,the further research in the future can be carried out for the automatic acquisition of index data and the analysis of noise data by using ANP method.