Ajuste proporcional integral discreto em relação aos dois últimos erros

Abstract

The production of a product without defects, demands that all processes must be continuously monitored and adjusted, in order to reach the best possible control state. In this research, it was made a link between statistical process control and engineering process control in a univariate case. This merging was exemplified through out a study made at Vonpar Refrescos S/A, where the pre-molding blowing machine was the main object of study. The statistical process control was achieved using controls charts as X-bar and R, in order to evaluate the stability of the process. The engineering control process was reached, using the Integral Proportional Control, related to the last two errors. To accomplish this, the multiple linear regression analysis and Exponentially Weighted Moving Average have been applied, in order to find the forecast and the smooth constant to measure the behavior of the residuals. The g constant, also called system impact, was found by the manual of the equipment and with the engineering help. The proposed control equation for the univariate case which also considers the last two forecasted errors, has been implemented on a spreadsheet, allowing easily application and showing the level of adjustment that needs to be made. With the implementation of this application, a reduction of raw material costs for the company is made. Utilizing together these methodologies in industrial processes, one is expected to obtain an important tool to be applied in the improvement process for product quality.