Accelerated Recurrent Neural Network Dynamics for Time-Varying Lyapunov Equation Solving

Designing new accelerated families of recurrent neural networks (RNNs) is essential for addressing diverse engineering problems. However, an efficient activation function (AF) must be carefully designed to improve the convergence of RNNs. In this paper, a novel AF is introduced to accelerate the solution of the time-varying Lyapunov equation (TVLE) using RNNs. Convergence analysis and experimental evaluation using Simulink demonstrate the efficiency of the proposed dynamical system.