ABSTRACT

With the growing complexity of manufacturing systems nowadays, the effective assessment of important risk factors inherent in the manufacturing process is crucial for the stability and reliability of such complex systems. Thus, this article proposes a data-driven approach using the state-of-art machine learning techniques to assess and forecast the procurement risks of non-ferrous metals associated with complex manufacturing systems. A variety of state-of-art machine learning models including ANN, LSTM, BLSTM, GARCH, as well as their combinations which compose the proposed hybrid models, are deployed and analyzed. The testing results show that the proposed hybrid machine learning method can forecast the price uncertainty in procurement and effectively evaluate the procurement risk in a precautious manner. Moreover, it is shown that the hybrid model that combines GARCH, ANN, and LSTM significantly improves the forecasting results. Besides, the optimal choice of the network configurations in the hybrid model is also analyzed via a series of sensitivity analyses. This research can serve as a useful reference for the effective assessment and control of procurement risk for manufacturing firms.

摘要

随着当今制造系统日益复杂，对制造过程中固有的重要风险因素进行有效评估对于此类复杂系统的稳定性和可靠性至关重要。因此，本文提出了一种数据驱动的方法，使用最先进的机器学习技术来评估和预测与复杂制造系统相关的有色金属采购风险。部署和分析了各种最先进的机器学习模型，包括 ANN、LSTM、BLSTM、GARCH，以及构成所提出的混合模型的它们的组合。测试结果表明，所提出的混合机器学习方法可以预测采购中的价格不确定性，并以预防的方式有效地评估采购风险。而且，结果表明，结合 GARCH、ANN 和 LSTM 的混合模型显着提高了预测结果。此外，还通过一系列敏感性分析分析了混合模型中网络配置的最佳选择。本研究可为制造企业有效评估和控制采购风险提供有益的参考。