Data-driven Control And Intelligent Optimization Of Polyester Fiber Spinning Process

Polyester fiber is a kind of polymer material produced from esterification and polycondensation of polyhydric alcohols and polyatomic acid.Its excellent performances on physical,chemical and mechanical aspects,such as the anti-wrinkling and form-maintaining properties,lead to a wide range of applications on clothing,electronic and architectural industry and make polyester fiber become the largest-scale production among numerous types of synthetic fibers.The production of polyester fiber becomes a significant industry relating to the national economy developments and social civilizations.However,the production process of polyester fiber is such a complex course because of various chemical reactions and physical changes.It is necessary to improve the polyester flow in the aspects of technical renovation based on the existing production technology,building an accurate model to reflect real manufacturing process and designing corresponding controller in order to reduce the production cost and improve the quality of products,which are the key points of further healthy development of polyester industry.The main contributions of this paper are as follows:(1)For the sake of eliminating the effect caused by human factors on controller performances,an immune-based optimization algorithm is used to optimize four control law parameters of the model-free adaptive controller.A weighted performance index is determined based on ITAE.The response speed and overshoot are both refined by means of the adjustment of weights.In this case,instead of regulating four parameters of MFAC manually,we can realize different control performances by adjusting only one weight value.(2)By adjusting the four parameters of MFAC control law collectively,the experimental results in last chapter reflect that the simultaneous optimization of the four parameters cannot give attention to both the overshoot and response speed of the system at the same time.Furthermore,we find that the control performance is insensitive to two of these parameters.So we consider optimizing the other two remaining parameters,in other words,they are also the most two important parameters.Based on neural-endocrine-immune hormone feedback regulation mechanism,we operate an adaptive improvement of parameters,which makes the error information in the previous moment will affect the setting of current parameters of MFAC controller.Based on this improvement,the controlling parameters can get real-time adjustment.The experiments show that the system adjusts parameters according to the controlling error instead of manual adjustment,so as to reduceing the influence of human factors.Even while the control object changes during the period of controlling process,the system can adjust controlling parameters automatically and realize the real-time tracking control.(3)In view of the complex multi-input multi-output system of polyester fiber spinning process,the ELM learning algorithm is adopted to set up the data model of the influencing factors and performance indexes of the existing polyester fiber spinning process,so as to replace the complex mechanism modeling process.The data model is used to produce the input and output data of the production process.In this case,the multi-input and multi-output system of the polyester spinning process can be controlled without establishing mechanism model.Finally,according to the content discussed in this paper,the aspects which can be optimized and perspective research problems are put forward.