Showing 2 results for Autocorrelated Process
Mahdi Imanian, Aazam Ghassemi, Mahdi Karbasian,
Volume 31, Issue 1 (3-2020)
Abstract
This work used two methods for Monitoring and control of autocorrelated processes based on time series modeling. The first method was the simultaneous monitoring of common and assignable causes. This method included applying five steps of data gathering, normality test, autocorrelation test, model selection and control chart selection on all non-stationary process observations. The second method was a novel one for the separate monitoring and control of common and assignable causes. In this method, the process was divided into the parts with and without assignable causes.
The first method was greatly non-stationary due to not separating common and assignable causes. This method also implied that the common causes were hidden in the process. The novel method for the separate monitoring of common and assignable causes could turn the process into a stationary one, leading to identifying, monitoring, and controlling common causes without any interference from the assignable causes. The results showed that, unlike the first method, the second method could be very sensitive to the common causes; it could, therefore, suitably monitor, identify and control both assignable and common causes.
The current work was aimed to use control charts to monitor and control the bootomhole pressure during the drilling operation.
Samrad Jafarian-Namin, Mohammad Saber Fallahnezhad, Reza Tavakkoli-Moghaddam, Ali Salmasnia, Mohammad Hossein Abooei,
Volume 32, Issue 4 (12-2021)
Abstract
In recent years, it has been proven that integrating statistical process control, maintenance policy, and production can bring more benefits for the entire production systems. In the literature of triple-concept integrated models, it has generally been assumed that the observations are independent. However, the existence of correlated structures in some practical applications put the traditional control charts in trouble. The mixed EWMA-CUSUM (MEC) control chart and the ARMA control chart are effective tools to monitor the mean of autocorrelated processes. This paper proposes an integrated model subject to some constraints for determining the decision variables of triple concepts in the presence of autocorrelated data. Three types of autocorrelated processes are investigated to study their effects on the results. Moreover, the results of the MEC and ARMA charts are compared. Due to the complexity of the model, a particle swarm optimization (PSO) algorithm is applied to select optimal decision variables. An industrial example and extensive comparisons are provided
