Scholarlink

In this paper, a series-parallel differential neural network is utilized to identify the measurable dynamics of a supply chain. A Lyapunov-like analysis, based on the Corollary to Barbalat’s Lemma, is employed to achieve three key objectives: 1) deduce a differential learning law for online updating of a synaptic weight matrix, 2) guarantee the boundedness of the synaptic weights, and 3) ensure the asymptotic convergence to zero of the identification error. Notably, the requirement for a Riccati or Lyapunov equation is effectively circumvented, consequently simplifying the tuning process considerably. The neural identifier is tested via simulation, utilizing a first-principles model of a supply chain solely as a data generator, rather than a live system. The simulation results demonstrate satisfactory performance.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Identification of Supply Chains Using Differential Neural Networks

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • José Humberto Pérez Cruz,
  • Alin Michel Zarate Cuapio,
  • Yael Antonio Otero Ritter,
  • Salvador Tavera Mosqueda

摘要

In this paper, a series-parallel differential neural network is utilized to identify the measurable dynamics of a supply chain. A Lyapunov-like analysis, based on the Corollary to Barbalat’s Lemma, is employed to achieve three key objectives: 1) deduce a differential learning law for online updating of a synaptic weight matrix, 2) guarantee the boundedness of the synaptic weights, and 3) ensure the asymptotic convergence to zero of the identification error. Notably, the requirement for a Riccati or Lyapunov equation is effectively circumvented, consequently simplifying the tuning process considerably. The neural identifier is tested via simulation, utilizing a first-principles model of a supply chain solely as a data generator, rather than a live system. The simulation results demonstrate satisfactory performance.