This paper focuses on synchronization for stochastic semi-Markov jump neural networks with time-varying delay via dynamic event-triggered scheme. The neural networks under consideration are described by It$\widehat{o}$ stochastic differential equations with semi-Markov jump parameters. First, supplementary variable technique and plant transformation are adopted to convert a phase-type semi-Markov process into an associated Markov process. Second, through stochastic analysis method and LaSalle-type invariance principle, novel sufficient conditions are deduced to realize stochastic synchronization for semi-Markov jump neural networks. Third, less conservative results are obtained compared with the existing methods. Finally, an industrial four-barrel model is applied to validate the superiority of the main algorithm.

本文重点研究具有时变延迟的随机半马尔可夫跳跃神经网络通过动态事件触发方案的同步。所考虑的神经网络由 It 描述$\widehat{○}$具有半马尔可夫跳跃参数的随机微分方程。首先，采用补充变量技术和植物变换将相型半马尔可夫过程转化为关联马尔可夫过程。其次，通过随机分析方法和拉萨尔型不变性原理，推导出实现半马尔可夫跳跃神经网络随机同步的新充分条件。第三，与现有方法相比，获得的结果不太保守。最后，应用工业四桶模型验证了主要算法的优越性。