Predictive Modeling for Printing Ink Consumption

In the current printing industries, precise forecasting of printing ink patterns is the key to cost reduction, inventory control, and environmentally friendly functioning. Conventional methods of estimation are based on coverage assumptions, which are always static and operator experience which frequently results in wastage of ink, delay in production and erratic quality. The paper provides an in-depth predictive modelling platform of ink consumption estimation based on statistical, machine learning, and deep learning methods. The proposed strategy is one that formulates ink usage prediction as a supervised regression, from which the heterogeneous inputs include the type of paper, the area covered, the color density, the print resolution, and the machine configuration parameters. The data is obtained during print job logs and in-built machine sensors and job specification files and past production logs. To increase predictive relevance and robustness, superior pre-processing methods are used, such as feature engineering of color coverage measures, ink density measures, and print complexity measures. The comparison between methods of baseline linear regression and statistical forecasting models and machine learning methods including decision trees, random forest, support vectors regression, and gradient boosting are made. Moreover, the deep learning models such as artificial neural networks, long short-term memory networks, and hybrid architectures are determined to obtain nonlinear relationships and temporal dependencies between printing workflows. Experimental evidence shows that ensemble and deep learning models are much more successful than the classical approaches, with lower error in prediction and overall generalization to a variety of print jobs

Artificial neural network · Deep learning · Generalization · Gradient boosting · Predictive modelling · Random forest · Support vector machine · Color Science and Applications · Material Properties and Processing · Nanomaterials and Printing Technologies

• Machine learning and deep learning

  Open Access•Christian Janiesch, Patrick Zschech et al.•Electronic Markets•2021

  Add to list