Research On Quality Rule Acquisition And Quality Prediction Of Parts Processing Based On Data Mining

"Made in China 2025" strategy has put forward new requirements for product quality control.As a component unit of products,it is of great significance to implement the "made in China 2025" strategy to explore the application of the new generation technology in the quality control of parts processing.With the development of information technology,manufacturing enterprises have accumulated a large number of data of parts processing quality,but they can not use these data effectively.Data mining technology is an important data processing technology in the field of artificial intelligence.It can discover the potential knowledge behind the data and solve the current situation that manufacturing enterprises have rich data on parts processing and poor knowledge on parts processing quality.First of all,in view of the parts processing quality data scattered in multiple business databases,inconvenient to carry out data mining directly,the parts processing quality data warehouse was established.The demand analysis,conceptual model,logical model and physical model of the part processing quality data warehouse were carried out,and the creation of the part processing quality data warehouse and its data ETL(data extraction,transformation and loading)process were realized.Then,aiming at the problem that the imbalance of enterprise quality data makes it difficult to find out the relevant rules of unqualified products,an acquisition model of parts processing quality rules based on improved mixed sampling and fp-tree is proposed.The mass data were equalized by using the hybrid sampling equalization method based on under-sampling of key rules and modified oversampling,and the quality rules were mined by using the fp-tree algorithm.Use a piston production enterprise's actual production data,and based on the FP-tree method,the method based on random sampling and mixed FP-tree,from two aspects of correctness rules number,through the contrast analysis of the model show that reducing the qualified class rules under the premise of can dig out more unqualified class rules,and proved the correctness of the rules are obtained.Then,in order to solve the problems of high complexity and low accuracy in most of the current part processing quality prediction algorithms,a part processing quality prediction model based on improved genetic bee colony algorithm(IGBCA)and support vector machine(SVM)was proposed.The proposed IGBCA algorithm is used to search and optimize the parameters of S VM to improve the prediction effect of the part processing quality prediction model.lt is concluded that the prediction model based on IGBCA-SVM has higher accuracy and shorter convergence time than the prediction model based on genetic algorithm(GA)and artificial bee colony algorithm(ABCA).Finally,on the basis of the previous study,the functional modules and hierarchical architecture design of the system were completed,and the part processing quality rules acquisition and quality prediction system based on data mining was developed by using Microsoft Visual Studio 2019 development platform,SQL Server 2017 database,and Anaconda3.