Title: A new risk evaluation method for supply chain based on convolution neural network
Authors: Huanle Han; Lianguang Mo
Addresses: Department of Tourism Management, Yellow River Conservancy Technical Institute, Kaifeng, 475000, China ' College of Management, Hunan City University, Yiyang, 413000, China
Abstract: In the traditional supply chain risk assessment methods, the significance of selecting evaluation indexes is low, which leads to the problems of low fitting and poor accuracy of risk assessment results. This paper proposes a new convolution neural network method to measuring the risk of supply chain. All risk factors in the supply chain are analysed to clarify the relationship between different factors. The determination of the overall risk assessment index is done by identifying the coordination risk, logistics risk, information risk, and capital risk. The determination of the individual risk index is done by identifying the cost-profit rate, product qualification rate, and order lead time of the supplier risk. The manufacturing cost, product development cycle, and product flexibility are determined in the manufacturer's risk. On this basis, the measurement index system of this link is designed, though the convolution neural network to measure this measurement index system. Through analysis the highest fitting degree of this method was about 94%, the highest true rate is about 0.91, the lowest false positive rate is about 0.09.
Keywords: convolution neural network; influencing factors; supply chain; risk evaluation; evaluation index.
DOI: 10.1504/IJASS.2024.140026
International Journal of Applied Systemic Studies, 2024 Vol.11 No.2, pp.121 - 137
Accepted: 20 Feb 2024
Published online: 15 Jul 2024 *