Risk Prediction of Digital Transformation of Manufacturing Supply Chain Based on Principal Component Analysis and Backpropagation Artificial Neural Network

Abstract

Digital transformation of manufacturing is a hot topic among strategic managers of manufacturing companies. The crux of digital transformation lies in the digitalization of manufacturing supply chain (MSC). However, the digital transformation of the MSC is highly uncertain, owing to the dynamic and complex changes of its nodes and structure in response to growing customer demand and fierce market competition. To propel the MSC digital transformation, it is crucial to effectively identify and predict the risk factors in the course of digital transformation. Therefore, this paper attempts to help manufacturing companies in China to successfully switch to a digital MSC. Firstly, the risk sources of the MSC digitization were identified, and complied into an evaluation index system for the digital transformation of the MSC. Next, the principal component analysis (PCA) was performed to reduce the dimension of the original data by revealing the three key principal components, and then the characteristic parameters of risk prediction are selected, so as to simplify the structure of neural network and improve the speed and efficiency of network training. On this basis, a backpropagation neural network (BPNN) was constructed for predicting the risks in MSC digitization. The results of training the model based on some data show that the proposed BPNN model has a good predictive effect. Furthermore, our model was compared with the traditional artificial neural network (ANN) model on a test set. The comparison demonstrates that our model achieved better effect than the traditional model in risk prediction. The results also show that the selected three principal components are reasonable, and the evaluation index system is valuable. The research results provide new insights to the smooth digital transformation of the MSC.