Hans Henrik Niemann

Passive fault tolerant control of a double inverted pendulum : a case study

(28/10/2019) Passive fault tolerant control of a double inverted pendulum a case study A passive fault tolerant control scheme is suggested, in which a nominal controller is augmented with an additional block, which guarantees stability and performance after the occurrence of a fault. The method is based on the YJBK parameterization, which requires the nominal controller to be implemented in observer based form. The proposed method is applied to a double inverted pendulum system, for which an H_inf controller has been designed and verified in a lab setup. In this case study, the fault is a degradation of the tacho loop.

Robust performance of systems with structured uncertainties in state space

Abstract This paper considers robust performance analysis and state feedback design for systems with time-varying parameter uncertainties. The notion of a strongly robust H ∞ performance criterion is introduced, and its applications in robust performance analysis and synthesis for nominally linear systems with time-varying uncertainties are discussed and compared with the constant scaled small gain criterion. It is shown that most robust performance analysis and synthesis problems under this strongly robust ∞ performance criterion can be transformed into linear matrix inequality problems, and can be solved through finite-dimensional convex programming. The results are in general less conservative than those using small gain type criteria.

Active Fault Diagnosis Based on Stochastic Tests

Active Fault Diagnosis Based on Stochastic Tests The focus of this paper is on stochastic change detection applied in connection with active fault diagnosis (AFD). An auxiliary input signal is applied in AFD. This signal injection in the system will in general allow us to obtain a fast change detection/isolation by considering the output or an error output from the system. The classical cumulative sum (CUSUM) test will be modified with respect to the AFD approach applied. The CUSUM method will be altered such that it will be able to detect a change in the signature from the auxiliary input signal in an (error) output signal. It will be shown how it is possible to apply both the gain and the phase change of the output signal in CUSUM tests. The method is demonstrated using an example.

Active fault diagnosis in closed-loop systems

Abstract Active fault diagnosis (AFD) of parametric faults is considered in connection with closed loop feedback systems. AFD involves auxiliary signals applied on the closed loop system. A fault signature matrix is introduced in connection with AFD and it is shown that if a limited number of faults can occur in the system, a fault separation in the fault signature matrix can be obtained. Then the single elements in the matrix only depend of a reduced number of parametric faults. This can directly be applied for fault isolation. If it is not possible to obtain this separation, it is shown how the fault signature matrix can be applied for a dynamical fault isolation, i.e. fault isolation based on the dynamic characteristic of the fault signature matrix as function of the different parametric faults.

Robust model predictive control of a wind turbine

In this work the problem of robust model predictive control (robust MPC) of a wind turbine in the full load region is considered. A minimax robust MPC approach is used to tackle the problem. Nonlinear dynamics of the wind turbine are derived by combining blade element momentum (BEM) theory and first principle modeling of the turbine flexible structure. Thereafter the nonlinear model is linearized using Taylor series expansion around system operating points. Operating points are determined by effective wind speed and an extended Kalman filter (EKF) is employed to estimate this. In addition, a new sensor is introduced in the EKF to give faster estimations. Wind speed estimation error is used to assess uncertainties in the linearized model. Significant uncertainties are considered to be in the gain of the system (B matrix of the state space model). Therefore this special structure of the uncertain system is employed and a norm-bounded uncertainty model is used to formulate a minimax model predictive control. The resulting optimization problem is simplified by semidefinite relaxation and the controller obtained is applied on a full complexity, high fidelity wind turbine model. Finally simulation results are presented. First a comparison between PI and robust MPC is given. Afterwards simulations are done for a realization of turbulent wind with uniform profile based on the IEC standard.

A model-based approach to fault-tolerant control

A model-based approach to fault-tolerant control A model-based controller architecture for Fault-Tolerant Control (FTC) is presented in this paper. The controller architecture is based on a general controller parameterization. The FTC architecture consists of two main parts, a Fault Detection and Isolation (FDI) part and a controller reconfiguration part. The theoretical basis for the architecture is given followed by an investigation of the single parts in the architecture. It is shown that the general controller parameterization is central in connection with both fault diagnosis as well as controller reconfiguration. Especially in relation to the controller reconfiguration part, the application of controller parameterization results in a systematic technique for switching between different controllers. This also allows controller switching using different sets of actuators and sensors.

Active fault diagnosis in closed-loop uncertain systems

Abstract Fault diagnosis of parametric faults in closed-loop uncertain systems by using an auxiliary input vector is considered in this paper, i.e. active fault diagnosis (AFD). The active fault diagnosis is based directly on the socalled fault signature matrix , related to the YJBK (Youla, Jabr, Bongiorno and Kucera) parameterization. Conditions are given for exact detection and isolation of parametric faults in closed-loop uncertain systems.

A μ-synthesis approach to robust control of a wind turbine

The problem of robust control of a wind turbine is considered in this paper. A set of controllers are designed based on a 2 degrees of freedom linearized model of a wind turbine. An extended Kalman filter is used to estimate effective wind speed and the estimated wind speed is used to find the operating point of the wind turbine. Due to imprecise wind speed estimation, uncertainty in the obtained linear model is considered. Uncertainties in the drivetrain stiffness and damping parameters are also considered as these values are lumped parameters of a distributed system and therefore they include inherent uncertainties. We include these uncertainties as parametric uncertainties in the model and design robust controllers using the DK-iteration method. Based on estimated wind speed a pair of controllers are chosen and convex combination of their outputs is applied to the plant. The resulting set of controllers is applied on a full complexity simulation model and simulations are performed for stochastic wind speed according to relevant IEC standard.

Active fault diagnosis by controller modification

Two active fault diagnosis methods for additive or parametric faults are proposed. Both methods are based on controller reconfiguration rather than on requiring an exogenous excitation signal, as it is otherwise common in active fault diagnosis. For the first method, it is assumed that the system considered is controlled by an observer-based controller. The method is then based on a number of alternate observers, each designed to be sensitive to one or more additive faults. Periodically, the observer part of the controller is changed into the sequence of fault sensitive observers. This is done in a way that guarantees the continuity of transition and global stability using a recent result on observer parameterization. An illustrative example inspired by a field study of a drag racing vehicle is given. For the second method, an active fault diagnosis method for parametric faults is proposed. The method periodically adds a term to the controller that for a short period of time renders the system unstable if a fault has occurred, which facilitates rapid fault detection. An illustrative example is given.

Model predictive control of wind turbines using uncertain LIDAR measurements

The problem of Model predictive control (MPC) of wind turbines using uncertain LIDAR (LIght Detection And Ranging) measurements is considered. A nonlinear dynamical model of the wind turbine is obtained. We linearize the obtained nonlinear model for different operating points, which are determined by the effective wind speed on the rotor disc. We take the wind speed as a scheduling variable. The wind speed is measurable ahead of the turbine using LIDARs, therefore, the scheduling variable is known for the entire prediction horizon. By taking the advantage of having future values of the scheduling variable, we simplify state prediction for the MPC. Consequently, the control problem of the nonlinear system is simplified into a quadratic programming. We consider uncertainty in the wind propagation time, which is the traveling time of wind from the LIDAR measurement point to the rotor. An algorithm based on wind speed estimation and measurements from the LIDAR is devised to find an estimate of the delay and compensate for it before it is used in the controller. Comparisons between the MPC with error compensation, the MPC without error compensation and an MPC with re-linearization at each sample point based on wind speed estimation are given. It is shown that with appropriate signal processing techniques, LIDAR measurements improve the performance of the wind turbine controller.