Fault Detection, Isolation And Prognosis For Control Moment Gyros Onboard Satellites

Developing a Diagnosis, Prognosis and Health Monitoring (DPHM) framework for a small satellite is a challenging task due to the limited availability of onboard health monitoring sensors and computational budget. This thesis deals with the problem of developing DPHM framework for a satellite attitude actuator system that uses a single gimballed Control Moment Gyro (CMG) in pyramid configuration as an actuator. This includes the development of computationally light data-driven model, fault detection, isolation and prognosis algorithms that works only using the attitude rate measurements from the satellite. A novel scheme is proposed for developing a data-driven model which fuses the symmetric property of the data and the system orientation property of actuators that reduces the need for historical data by 93.75%. The data is trained using Chebyshev Neural Network. A threshold based fault detection algorithm is used to detect the faults of spin motor and gimbal motor used in a CMG. A novel optimization based fault isolation formulation is developed and simulated for given uniformly distributed system parameters. The algorithm has a success rate of 93.5% in isolating faults of 8 motors (4 gimbal and 4 spin) that can fail in 254 different ways. For Fault Prognosis, an error based scheme is developed as a measure of degradation. General path model with Bayesian updating is used for predicting the remaining useful life of the spin motor. It performs with 96.25% accuracy when 30% of data is available. Overall, the proposed algorithms can be regarded as a promising DPHM tool for similar non-linear systems.

Fault Detection, Isolation And Prognosis For Control Moment Gyros Onboard Satellites

Developing a Diagnosis, Prognosis and Health Monitoring (DPHM) framework for a small satellite is a challenging task due to the limited availability of onboard health monitoring sensors and computational budget. This thesis deals with the problem of developing DPHM framework for a satellite attitude actuator system that uses a single gimballed Control Moment Gyro (CMG) in pyramid configuration as an actuator. This includes the development of computationally light data-driven model, fault detection, isolation and prognosis algorithms that works only using the attitude rate measurements from the satellite. A novel scheme is proposed for developing a data-driven model which fuses the symmetric property of the data and the system orientation property of actuators that reduces the need for historical data by 93.75%. The data is trained using Chebyshev Neural Network. A threshold based fault detection algorithm is used to detect the faults of spin motor and gimbal motor used in a CMG. A novel optimization based fault isolation formulation is developed and simulated for given uniformly distributed system parameters. The algorithm has a success rate of 93.5% in isolating faults of 8 motors (4 gimbal and 4 spin) that can fail in 254 different ways. For Fault Prognosis, an error based scheme is developed as a measure of degradation. General path model with Bayesian updating is used for predicting the remaining useful life of the spin motor. It performs with 96.25% accuracy when 30% of data is available. Overall, the proposed algorithms can be regarded as a promising DPHM tool for similar non-linear systems.

Adaptive Neural Backstepping Control of Variable Stiffness Actuator

In this paper, we provide a novel adaptive neural network backstepping control scheme for a special variable stiffness actuator (VSA) based on lever mechanisms with saturation inputs, output constraints and disturbances is presented here. In the controller designing, the prescribed performance-tangent barrier Lyapunov function (PP-TBLF) is introduced to ensure that both the prescribed performance bound of tracking error and the output constraints are not violated. In specific steps of backstepping control scheme, the Chebyshev neural network and the Nussbaum-type function are used to solve the unknown nonlinearities and unknown gain sign. Meanwhile, the inverse hyperbolic sine function tracking differentiator is exploited to solve the “explosion of complexity” caused by the differentiation of virtual inputs and also approximate the complex partial derivative caused by the auxiliary control signals. Finally, the stability of the whole scheme is proved by Lyapunov criterion and the simulation results are presented to illustrate the feasibility of the raised control strategy.

Adaptive Chebyshev Neural Network Control for Ventilator Model under the Complex Mine Environment

Ventilator is important equipment for mines as it safeguards the lives under the shaft and ensures other equipment’s proper functioning by providing fresh air. Therefore, how to effectively control the ventilator system becomes more significant. In order to acquire the commonly used model and control strategy for ventilator systems, a new universal ventilator model is established based on the blast capacity differential pressure in the ventilating duct and the ventilator motor model. Then, an adaptive Chebyshev neural network (ACNN) controller is proposed to effectively control the ventilator system where the unknown load torque and the unknown disturbance caused by the complex environment under the shaft are approximated by the Chebyshev neural network (CNN). Afterwards, an appropriate Lyapunov function candidate is designed to guarantee the stability of the proposed controller and the closed-loop ventilator system. Finally, the ACNN controller has been demonstrated to be effective in terms of validity and precision for the new proposed ventilator model through the simulations.

Solving two-dimensional linear partial differential equations based on Chebyshev neural network with extreme learning machine algorithm

Purpose This paper aims to develop a novel algorithm and apply it to solve two-dimensional linear partial differential equations (PDEs). The proposed method is based on Chebyshev neural network and extreme learning machine (ELM) called Chebyshev extreme learning machine (Ch-ELM) method. Design/methodology/approach The network used in the proposed method is a single hidden layer feedforward neural network. The Kronecker product of two Chebyshev polynomials is used as basis function. The weights from the input layer to the hidden layer are fixed value 1. The weights from the hidden layer to the output layer can be obtained by using ELM algorithm to solve the linear equations established by PDEs and its definite conditions. Findings To verify the effectiveness of the proposed method, two-dimensional linear PDEs are selected and its numerical solutions are obtained by using the proposed method. The effectiveness of the proposed method is illustrated by comparing with the analytical solutions, and its superiority is illustrated by comparing with other existing algorithms. Originality/value Ch-ELM algorithm for solving two-dimensional linear PDEs is proposed. The algorithm has fast execution speed and high numerical accuracy.