A New Reliability Prediction Model in Manufacturing Systems

Reliability prediction has been widely studied in many research fields to improve product and system reliability in manufacturing systems. Traditionally, to establish the prediction model, modelers would use all training data without preference. However, the prediction model based only on the most recent data may have better performance. In this paper, to realize an accurate prediction with the most recent data sets, we use the grey model to establish the reliability model. Then, the cubic spline function is integrated into the grey model to enhance the prediction capability of GM(1, 1), a single variable first order grey model. The newly generated model is defined as 3spGM(1, 1). To further improve the prediction accuracy, the particle swarm optimization (PSO) algorithm is applied to 3spGM(1, 1). We call the improved version P-3spGM(1, 1). Finally, we validated the effectiveness of the proposed model using failure data sets of electric product manufacturing systems.