A BP-SVM combined model for intermittent spare parts demand prediction

Accurate and timely prediction of demand plays an important role in the process of spare parts procurement and inventory management. At the same time, it is difficult to predict the demand of some spare parts due to the strong randomness of demand occurrence and the obvious discontinuity of demand distribution. In recent years, many different prediction methods have been proposed, but there are still some limitations. Therefore, there is still a lack of mature intermittent demand forecasting methods in academia. This paper takes one of the spare parts with intermittent demand in a power plant as an example. On the basis of considering the influencing factors of spare parts demand, this paper separates the demand interval and demand quantity of the intermittent demand, and proposes a combined prediction model of BP-SVM to predict the demand interval and quantity respectively. In this paper, the prediction results of the BP-SVM combined model are compared with those of BP neural network model and support vector machine. The results show that the BP-SVM combined model in this paper has better forecasting performance than BP and SVM. It is also proved that the BP-SVM combined model proposed in this paper is suitable for the prediction of spare parts with intermittent demand.