New Noise-Tolerant ZNN Models With Predefined-Time Convergence for Time-Variant Sylvester Equation Solving

Sylvester equation is often applied to various fields, such as mathematics and control systems due its importance. Zeroing neural network (ZNN), a systematic design method for time-variant problems, has been proved be effective on solving in the ideal conditions. In this paper, order realize predefined-time convergence of ZNN model modify robustness, two new noise-tolerant ZNNs (NNTZNNs) are established by devising novelly constructed nonlinear activation functions (AFs) find accurate solution presence noises. Unlike original models activated known AFs, proposed NNTZNN novel therefore, possessing excellent strong robustness even Besides, detailed theoretical analyses ability given considering different kinds Simulation comparative results further verify performance models, when online equation.