Research On Abnormal Diagnosis Of Process Quality Based On CS-FSVM

Quality control of the manufacturing process is the key to ensure product quality.How to effectively detect and diagnose anomalies is a practical issue that needs to be solved for multiple quality features that are related to each other.However,the traditional methods for quality diagnosis have many shortcomings.In the face of complex and high-tech products,the diagnostic requirements are often not met.Based on this,on the basis of reviewing and summarizing related theories of quality diagnosis and its research,this paper conducts a general analysis of related research with the software of Cite Space.Taking the multivariate process quality in the manufacturing process as the research object.In view of the fact that the product is influenced by several related key quality characteristics in the manufacturing process,the multivariable control chart cannot identify the specific abnormal variable and the difficulty of diagnosis increases.This paper uses the MEWMA control chart and constructs a fuzzy support vector machine(FSVM)model to achieve multiple The diagnosis of variable process quality.The main research content includes:(1)FSVM model construction and parameter optimization.Considering that traditional support vector machines(SVMs)are sensitive to sample noise when classifying abnormal variables to reduce the diagnostic accuracy,the method of calculating the sample membership can reduce the influence of sample noise.However,the commonly used method of calculating the membership degree is easy to ignore the sample.Because of the ambiguity and randomness of the cloud model,the cloud model is used to calculate the degree of membership of the sample points in each anomaly category and to measure the ambiguity and randomness of the sample.A fuzzy support vector machine model is constructed..In order to further improve the efficiency of diagnosis,this paper uses the global optimal Optimum Search(CS)algorithm to optimize the model’s penalty parameter C and kernel function parameter g,and introduces the optimization process.(2)Model testing and multivariate process quality diagnostics.The process of multivariate process quality mean shift diagnosis and the seven types of abnormal patterns of mean shift of quality characteristics were analyzed.The sample data of each abnormal pattern was simulated according to the simulation method of mean shift anomaly,and the sample was constructed by constructing the MEWMA control chart.Data monitoring and analysis,the use of the sample data to calculate the degree of membership and parameter optimization simulation analysis,and model training and testing,to achieve the classification of the model 7 classification of abnormalities,the test results show that the accuracy of diagnosis was 97.4286%,and improvement Compared with the pre-optimization model,the constructed model has a higher diagnostic accuracy.(3)Example application analysis.To illustrate the validity and practicality of the model,the CS-FSVM model was applied to the manufacturing process of the capacitor products of the example company for case analysis.The sample data of two key quality characteristics of core diameter and core height were collected to diagnose The process performed anomaly diagnosis on two quality features.The results showed that the final diagnostic accuracy rate was 95.603%.It was verified that the constructed CS-FSVM model can be effectively applied to the multivariable process quality anomaly diagnosis with good diagnostic performance.