Research On Quality Control Of Knitting Products Based On Data Mining

As the country's "San Pin" strategy is continuously implemented in all walks of life,the " San Pin" strategy has been formally included in the "13th Five-Year Plan" of the textile industry,requiring textile enterprises to pay more attention to product quality issues.In addition,market competition has become increasingly fierce,and the market competition among knitting companies has become more of a competition for product quality.Product quality is increasingly becoming a decisive factor in whether companies can establish themselves in fierce market competition.However,most knitting companies are still using traditional statistical process control(SPC)methods to control product quality during the production process.This method is old-fashioned and has many drawbacks.In the new era of higher quality requirements for knitting enterprises,it will be overstretched.Based on this,this paper applies the data mining technology to the research of the quality control in the production process of knitting enterprises.Through data mining research on a large amount of historical data of product quality in enterprise's MES,a new quality control method was proposed to provide reliable and effective prior decision-making for companies to control product quality in the production process and further improve product quality.First of all,through the in-depth study of related technologies such as data mining's principles,technology methods,software tools and so on,this paper establishes the methods and steps of data mining research,which lays a good foundation of theoretical knowledge for the establishment of quality control model.By analyzing the quality control methods of production processes that are commonly used at present,the drawbacks of the SPC methods are pointed out,and the predetermined target of this data mining research are determined.According to the intended objectives of this study,comparing and analyzing the characteristics of commonly used data mining algorithms,choose the most suitable decision tree C5.0 algorithm to build the quality control decision tree model,and use the program language of Python to implement algorithmic programming.Secondly,according to the principle of the decision tree C5.0 algorithm,the construction process of the quality control model for the knitting product production process is determined.Through analysising of the general production process of knitting products,the main influencing factors of product quality in the production process are determined,including product type,density,raw material supplier,raw material quality grade,machine,machine speed,operator,environmental temperature and humidity,and shifts.The selected influencing factors are taken as the test attributes of the model,and the “product quality level” index is taken as the target attribute.Based on the principle of the C5.0 algorithm,a preliminary decision tree model is constructed.Then the Post-pruning method is used to prune the preliminary model,which effectively reduced the degree of over-fitting.The basic Boosting algorithm is used to further optimize the model,which significantly improves the accuracy of predictive classification results and effectively improves the overall performance of the model.Finally,according to the experimental results of the series of algorithms,a final quality control decision tree model is generated.Finally,based on the construction process of the quality control model,the model is demonstrated.The historical data of the product quality is selected from the MES database of a knitting company,and it is cleaned,filled,transformed to obtain a target data set that can be used for data mining.The program language of Python is used to complete the modeling algorithm programming and process the target data to obtain a quality control decision tree.The model is tested and evaluated from both the accuracy and stability of the predictive classification,which proves that the obtained model has high accuracy,good stability and high practical application value.Finally,the classification rule set of the quality control decision tree is extracted,and the application of the decision result in the actual quality control of the knitting product production process is analyzed.According to the results of the decision-making arrangement of production practices,the overall ratio of A and other products of the company's weaving fabrics increased from 93% to about 98%.The quality control method of knitting products based on data mining in this paper remedies the defects of traditional SPC methods,to provide reliable and effective prior decision-making of quality control in the production process for knitting enterprises,and achieve better control of the products' quality.