Research And Application On Method Of Reconfigurable Textile Intelligent Process Planning And Virtual Machining

The international textile market’s competition of quality,efficiency and price is essentially the competition of high and new technologies employed to reform the traditional textile industry,and also is the competition of the scientific innovation ability.As the key link of textile production process,textile process planning’s ability to agile response to the changing demands of the market directly affected the core competitiveness of textile enterprises.Essentially,various complicated knowledge is comprehensive used to textile process decision-making and reasoning,due to the inherent characteristics of textile production,such as long flow process,many nonlinear coupling variable,and complicated machining mechanism etc.,for a long time,by means of the traditional process planning,it is difficult to resolve the problems such as how to acquire and reuse process knowledge,how to improve efficiency and quality of process planning,and how to predict and control product quality effectively.In order to raise the level of product development and manufacturing,agile response to the changing market requirements,it is needed urgently to reform the traditional means of process planning with advanced information technologies and intelligent technologies in textile industry.With the help of fund of national technology innovation project and national technology support plan project,the theory and method of intelligent textile process planning and virtual machining system(ITPP&VMS)are discussed in the paper.An intelligent method of textile process planning based on hybrid knowledge reuse is proposed,and a series of methods about quality predict and control which support virtual machining are discussed in details in the paper.The main research contents of the paper are summarized as follows:(1)A reconfigurable design methodology of distributed textile CAPP system.The function requirements,fundamental structure,systematic business procedure and intelligent characters of ITPP&VMS are analyzed in detail,and the system processes consist of the functions,including process planning,quality forecast and control,are explained in an integrated system flow chart.On the basis of analysis above,a service-oriented architecture for ITPP&VMS is designed,and the principles of process knowledge reuse service are discussed.In order to achieve transformation of relational data model to process knowledge service object model in the n-tiered system architecture,three different approaches to representing an process object inheritance hierarchy are illustrated respectively by using the ADO.NET entity framework,and the application effects of different mapping schemes are compared according to the different usage scenarios.In order to improve the system’s ability to response quickly to the changing business process,a dynamic assembled component model based on managed extensible framework(MEF)is given to support reconfigurable virtual machining in the paper.(2)The methods of textile process knowledge acquisition and reuse.Based on ontology,the ontological structure of process knowledge is presented in the paper;both OWL description and its conversion to relationship database are discussed to get the sharing conceptual model for textile process planning system.As to intelligent process planning methods,the similarity algorithms and steps for case-based textile process reasoning are discussed;the model of rule guiding based reasoning(RGBR)is proposed and applied to design spinning process parameter.Based on RGBR,relationship database representation of textile process rules is explained.In order to take advantage of each intelligent method stated in the article,the integrated process planning approaches based on hybrid knowledge representation and reasoning is proposed to realize process planning in progressive manner.As far as knowledge discovery and acquisition is concerned,a textile process decision rules extracting method based on rough set theory(RST)is studied,and the key algorithms that involves data discretion,attribute reduction and rule reduction are also introduced.In order to discover the law of influence on the yarn strength when fibers and corresponding spinning variables are changed,a rule extraction experiment with actual production dataset is designed,the logical IF-THEN rules extracted from the decision table indicate that the original strength of fibers is a key factor influencing on the yarn strength,that the different values combination of the final reduced attributes also obviously influence on the yarn strength in different degree.Therefore,RST method can be taken into account for spinners to acquire textile process knowledge and design the quality forecast model.(3)The methodology of quality prediction and process planning optimization for textile virtual machining.The method of textile virtual machining characterized mainly by quality prediction is studied.In order to provide theoretical basis for the reasonable designing,evaluating and optimizing of quality prediction model,as well as the decision-making and optimizing of process,the model of virtual machining performance evaluating and decision-making(VMED)is proposed in the article.Based on intelligent methods of back-propagation neural network(BPNN)and support vector regression machines(SVR),prediction models for different spinning quality characteristics are designed and model parameters are optimized by the means of genetic algorithm.Using the sample dataset from real production,experiments are set up and the resulting prediction accuracy of different model with different parameters is analyzed and compared to validate the model performance.The experimental results show that both BPNN-based and SVR-based quality prediction are suitable to textile production prediction,and it is evidently that predict accuracy of most quality characteristics in spinning process can reach 80～95%.In contrast to BPNN-based model,SVR-based model is more suitable to small samples machine-learning and is capable of good generalization.Furthermore,the application fields of textile virtual machining is broaden by combining.the technologies of quality prediction and CBR-based process planning to evaluate process schema and optimize process parameters.(4)Quality statistical process control methods for textile customization production.On the basis of quality control requirements analysis in textile process planning system,an approach to quality statistical process control(SPC),which had been integrated with knowledge acquisition,rule-based compensation of process measure,fault pattern recognition,is proposed in the paper.Through the statistical analysis of the actual production quality data,the methods of the textile manufacturing process capability assessment,process stability diagnosis and control charts are illustrated.Due to some limitations in traditional SPC technologies,a control chart pattern(C.CP)recognition method based on support vector classification machines(SVC)is studied especially.According to the dataset of six kinds of CCP generated by using Monte Carlo,SVC-CCP model and its experiments are designed,the results shows that the pattern recognition accuracy of SVC-CCP model is up to 90%more.Compared with the SVC-CPP model based on observed values,SVC-CCP model based on time domain features is more suitable to pattern recognition of small sample in customized textile production.Actually,two methods can be combined to meet the demands of textile quality control effectively.(5)Realization and application of intelligent textile process planning and virtual machining.Application to cotton and wool textile industry as background,the software named as web-based intelligent textile process planning and virtual machining(WITPVM)is developed.The system structure and distributed features of WITPVM are introduced;the main functions such as intelligent process planning,quality prediction and control are illustrated with rich application instances of user interface.Based on service-oriented system architecture,two realizing technologies,including process planning service,process information representation and conversion based on XML,are expound to support the viewpoints that system can be customized,scalable。and reconfigurable.Engineering application case indicates that the software play a positive role in guiding textile production practice for process planning,quality prediction and control.Totally,because textile process planning has the characteristics of comprehensiveness,empirical and creativeness,it is very difficult to adapt the traditional approaches of process planning to fierce market competition,therefore,the article present the methods of intelligent textile process planning and virtual machining.Using the advanced information and intelligence technologies,the thesis study the textile CAPP platform level and common technologies which can be used to solving the hard problems existed in textile process planning,such as quality prediction and control,in order to improve the existing production process,to enhance new product development capacity and to ensure product quality.Both in theory and in practice,further research and promotion of application achievement are very important to enable textile designing and manufacturing to transform from"experience-depended" to "knowledge-driven".as well as to promote international competition of our textile product.