development of clustering technique and genetic algorithm to monitor multivariate descriptive processes based on large-scale nominal contingency tables (case study: renewable energy process )

Many real-world issues are based on multivariate processes with descriptive characteristics that are represented by contingency tables. a contingency table is a tool for showing the simultaneous relationship of two or more descriptive variables that is modeled by the log-linear communication function and monitored over time. in some statistical process monitoring (spm) applications, we are faced with the multiplicity of variables and, of course, the number of nominal classifications of the response variable. to model them, a log-linear model based on large-scale contingency tables is used that are called nominal large-scale descriptive multivariate processes. in monitoring this type of process, we face the negative impact of large dimensions of contingency tables on the performance of control charts. for this purpose, a new approach based on the clustering approach in correspondence analysis have been developed to reduce the effect of large dimensions and improvement performance of the control charts in diagnosing out of control status. the performance of control charts has been evaluated using simulated studies and the results indicate the appropriate efficiency of the proposed approach in reducing the impact of the contingency table dimensions on the performance of the control charts. in addition, to demonstrate the performance efficiency of the proposed methods, a real case study in the field of renewable energy has been used, the results of which indicate the proper performance of the proposed control charts in diagnosing out of control status.