Norm-Based Adaptive Coefficient ZNN for Solving the Time-Dependent Algebraic Riccati Equation
Chengze Jiang, Xiuchun Xiao
Abstract
The time-dependent algebraic Riccati equation (TDARE) problem is applied to many optimal control industrial applications. It is susceptible to interference from measurement noises in the virtual environment, which current methods cannot effectively address. A norm-based adaptive coefficient zeroing neural network (NACZNN) model to solve the TDARE problem is proposed, with an adaptive scale coefficient based on the residual error norm to accelerate convergence speed to the theoretical solution. Momentum enhancement terms enable NACZNN to effectively solve the TDARE problem in real time when perturbed by measurement noise. Simulation experiments were designed and executed, and results confirm the NACZNN model's superior robustness and accuracy when solving the TDARE problem disturbed by noises in real time.