Análise discriminante e otimização aplicadas ao controle de qualidade de ruído de compressores usando os softwares r e gurobi

Abstract

This work approaches the application of multivariate analysis techniques for quality control compressors noise levels using the software R. Due to the high competitiveness of the market in terms of technological change, internet, development of communications and transportation, globalization, among other factors, the quality of products has become an important competitive advantage in the industrial setting. Thus, this technique has wide employability in quality control, is the forecasting optimization weather, or agricultural experiments or as in this case study, in quality control compressors noise levels. For the simulation, we used the statistical analysis software R. This report describes the statistical theoretical development and simulation. In agreement with the initial proposal of the work, we used the calculation of the main vibroacoustic symptoms diagnosed by Duarte (2013): kurtosis, skewness, crest factor, K4, energy level of the filtered envelope and difference between the maximum and minimum of the filtered envelope. The symptoms were calculated and used as data entry in the multivariate model with the objective of determining Fisher's Linear Discriminant Function (FDLF). One of the results is, from the data of a chosen compressor, to determine if it is adequate or not, through application of the FDLF found. Another result is the application of a program in the Gurobi software, with academic license, object oriented in C ++ with dynamic allocation of memory to classify the compressors.