基于BP神经网络和蜂群算法对在T-JIT环境下供应链协同风险的预警研究  

Study on Early Warning of Supply Chain Coordination Risk in T-JIT Environment Based on BP Neural Network and Bee Colony Algorithm

作　　者：郝丽[1] 胡大伟[1] HAO Li;HU Da-wei(School of Automobile, Chang＇an University, Xi＇an Shaanxi 710064, China)

机构地区：[1]长安大学汽车学院,陕西西安70064

出　　处：《公路交通科技》2018年第6期112-120,共9页Journal of Highway and Transportation Research and Development

基　　金：中央高校基本科研业务费专项基金项目(310822151030)

摘　　要：为了保证T-JIT环境下供应链采购管理中顾客对产品多样化、小批量、准时化的发展需求和协同管理智能化、信息化的发展趋势,同时促进供应链采购管理、库存管理之间和供应商之间的协同稳定、高效发展,针对供应链上各节点不确定性环境的影响和需求复杂性问题的困扰,同时扩充供应链风险预警管理的理论方法和体系,提出了一种基于BP神经网络与蜂群算法(ABC-BP)相结合的优化元启发式算法,对供应链各节点协同风险进行预警分析,用蜂群算法解决了选取BP神经网络权值和阈值的随机性,采用BP神经网络对每组供应链进行协同风险预警区间划分,找出影响供应链协同风险的主要因素,随后采取相应措施预防风险发生,并用实例证明该方法的有效性与可行性。传统的BP神经网络存在收敛速度慢,容易陷入局部最小值等缺点,影响风险预警的准确性和可行性。改进的ABC-BP算法与以往BP神经算法相比时效性高、稳定性强、准确率高。该问题的研究可以预防和缓解供应链协同风险的发生,减少财力和物力的损失;同时可以满足顾客多品种、小批量、多样化的采购需求,保证供应链各节点企业之间的风险协同,为T-JIT环境下供应链协同中不确定性风险的预警管理提供了一种新的可行方法。In order to guarantee the customers development demand for product diversification, sma]l quantities, timely and the trends of inte]ligence and informatization under the collaborative procurement management of supply chain in T-JIT environment, and to promote eollaborative stability and efficient development among supply chain procurement management, inventory management and suppliers, aiming at each node in the supply chain is plagued by the uncertain environment and the complexity of demand and to expand the theoretical method and system of risk warning management in supply chain, an optimal meta- heuristics algorithm based on BP neural network and bee colony algorithm is proposed to analyze the collaborative risk of each node of supply chain. The randomness of the weights and thresholds of BP neural networks are solved by bee colony algorithm. The collaborative risk early warning intervals of each supply chain are divided by using BP neural network to find out the main factors influencing the coordination risk of the supply chain. Then, the corresponding measures are taken to prevent the risk occurrence, and the effectiveness and feasibility of the method are proved by an example. The traditional BP neural network has the disadvantages of slow convergence and easy to fall into local minimum, which affects the accuracy and feasibility of risk early warning. The improved ABC-BP algorithm has higher time efficiency, higher stability, and higher accuracy than previous BP neural algorithms. The study of this problem could prevent and mitigate the occurrence of supply chain collaborative risks, reduce the financial and material losses, and ensure the risk coordination among the enterprises at each node in the supply chain to fit the customer＇s multi-variety, small-batch, and diversified procurement requirements. It provides a new feasible method for warning management of supply chain collaborative uncertainty risk in T-JIT environment.

关 键 词：交通工程 供应链协同风险预警 ABC-BPNN算法 风险评价 权值和阈值 

分 类 号：U491.123[交通运输工程—交通运输规划与管理]