Modified ZND with Immunity to Periodic Noises for Solving Time-Varying Nonlinear Equations: A Control Perspective

In this paper, a modified zeroing neurodynamics (MZND) model with immunity to periodic noises is proposed from a control perspective to address time-varying nonlinear equations (TVNE), a common issue in engineering applications. Grounded in the Lyapunov stability theory, this paper provides a thorough assessment of the performance of the MZND model. Through comparative analyses and simulations, the superior performance of the MZND model is validated against existing neurodynamics models. The MZND model stands out for its inherent immunity to periodic noises, a critical feature for enhancing robust in noisy real-world environments. The practical efficacy of the MZND model is exemplified by its application in 3D dynamic sound source location, demonstrating exceptional  performance in engineering settings.

本文从控制论视角出发，提出一种具备抗周期噪声能力的改进型归零神经动力学（modified zeroing neurodynamics, MZND）模型，以求解工程应用中普遍存在的时变非线性方程（time-varying nonlinear equations, TVNE）问题。本文基于李雅普诺夫稳定性理论，对MZND模型的性能开展了全面评估。通过对比分析与仿真实验，本文验证了MZND模型相较于现有神经动力学模型的优越性能。MZND模型的核心亮点在于其天生具备抗周期噪声的特性，这一关键特征可有效提升模型在含噪真实工程环境中的鲁棒性。本文将MZND模型应用于三维动态声源定位任务，以此验证其实际应用效能，结果表明该模型在工程场景中展现出了卓越的性能。