Abstract: A nonlinear tracking differentiator is constructed by introducing the tangent Sigmoid function in the neural network and the terminal attractor function. The asymptotic convergence of the designed tracking differentiator is proved by the complete stability theory. Also, the characteristics of the frequency domain are analyzed by the frequency sweep test and the parameter tuning rules are summarized. Compared with the previous literature, the structure of the tracking differentiator is further optimized, and the parameters of overlapping functions are integrated to make it easier to adjust. At the same time, the terminal attractor function is added to reduce the flutter caused by the high frequency signal, so that the noise suppression capability is stronger. The simulation results show that the designed tracking differentiator is capable of filtering and deriving the sinusoidal signals that contain noise interference, and has the advantages of fast response speed and high precision. Simultaneously, the designed tracking differentiator achieves fast tracking and generalized derivative’s approximating of the square wave signal with stronger capabilities of noise suppression.

Key words: tracking differentiator, tangent Sigmoid function, terminal attractor function, parameter tuning