Adaptive neural control of microgravity active vibration isolation system subject to output constraint and unexpected disturbance

Abstract

The objective of this investigation is to explore the issue of microgravity vibration isolation control within a space station, while considering constraints on output and unexpected disturbances. Drawing upon an analysis of the environment and operating conditions within the space station, the uncertainties resulting from variable payloads and unforeseen disturbances are depicted as combined disturbances. On this basis, an actor–critic neural network with energy consumption evaluation is used to estimate the combined disturbances and devise strategies for compensatory control. Furthermore, an adaptive robust term is devised to handle errors generated by the estimation process of the neural network. Additionally, to ensure that the output of the isolation system remains within specified bounds, an improved predefined performance function is introduced, thereby ensuring adherence to constraints regardless of initial conditions. Simulations conducted under different operational scenarios demonstrate the effectiveness of the proposed controller in achieving system stability. The results indicate a remarkable suppression of onboard vibration amplitude, amounting to approximately 60 dB within the frequency range of 2 to 100 Hz.