Pujapanda, Asim and Acharya, Sibashish Priyadarshi (2007) Identification of control chart patterns using neural networks. BTech thesis.
To produce products with consistent quality, manufacturing processes need to be closely monitored for any deviations in the process. Proper analysis of control charts that are used to determine the state of the process not only requires a thorough knowledge and understanding of the underlying distribution theories associated with control charts, but also the experience of an expert in decision making. The present work proposes a modified backpropagation neural network methodology to identify and interpret various patterns of variations that can occur in a manufacturing process. Control charts primarily in the form of X-bar chart are widely used to identify the situations when control actions will be needed for manufacturing systems. Various types of patterns are observed in control charts. Identification of these control chart patterns (CCPs) can provide clues to potential quality problems in the manufacturing process. Each type of control chart pattern has its own geometric shape and various related features can represent this shape. This project formulates Shewhart mean (X-bar) and range (R) control charts for diagnosis and interpretation by artificial neural networks. Neural networks are trained to discriminate between samples from probability distributions considered within control limits and those which have shifted in both location and variance. Neural networks are also trained to recognize samples and predict future points from processes which exhibit long term or cyclical drift. The advantages and disadvantages of neural control charts compared to traditional statistical process control are iscussed. In processes, the causes of variations may be categorized as chance (unassignable) causes and special (assignable) causes. The variations due to chance causes are inevitable, and difficult to detect and identify. On the other hand, the variations due to special causes prevent the process being a stable and predictable. Such variations should be determined effectively and eliminated from the process by taking the necessary corrective actions to maintain the process in control and improve the quality of the products as well. In this study, a multilayered neural network trained with a back propagation algorithm was applied to pattern recognition on control charts. The neural network was experimented on a set of generated data.
|Item Type:||Thesis (BTech)|
|Uncontrolled Keywords:||Neural networks, X-bar chart, CCPs|
|Subjects:||Engineering and Technology > Mechanical Engineering|
|Divisions:||Engineering and Technology > Department of Mechanical Engineering|
|Deposited By:||Hemanta Biswal|
|Deposited On:||27 Jun 2012 14:48|
|Last Modified:||27 Jun 2012 14:48|
|Supervisor(s):||Patel, K R|
Repository Staff Only: item control page