A Neural Network Approach in Reducing Offset Printing Spoilages on Solid Bleached Boards

Abstract

Offset printing is the process in which the image is offset from a plate to a rubber blanket, then following to the printing surface. With the combination of the lithography, which depends on the aversion of oil and water, this technique produces a flat accurate image representation. One of the leading problems for offset printing industries are the spoilages. Spoilages greatly reduce profits due to the unsold rejected products that were produced. With corporations innovating new ways to improve their operations, this study will use an artificial neural network to analyze and predict the spoilage output per job run based on the inputs, ink quality, machine grade, design complexity, and raw material quality. The platform that is used is MATLAB. This software has great neural network capabilities that are widely used around the globe by leading data scientists. Data is obtained in excel format from the company that the researcher has come in contact with. The company has agreed to share some data as long as some of the information such as suppliers and brands that are serviced are not disclosed. The undisclosed information is not significant for the neural network; therefore, the data is sufficient enough to work on. With the results gathered and analyzed from the neural network system, it can be concluded that this research has been able to predict the spoilage output of a job run to a certain extent due to limited data. Recommendations, however, include a better standardization for the inputs, and more data gathered for the database in order to obtain more accurate results.