Kai Rothenhagen

Doubly Fed Induction Generator Model-Based Sensor Fault Detection and Control Loop Reconfiguration

Fault tolerance is gaining interest as a means to increase the reliability and availability of distributed energy systems. In this paper, a voltage-oriented doubly fed induction generator, which is often used in wind turbines, is examined. Furthermore, current, voltage, and position sensor fault detection, isolation, and reconfiguration are presented. Machine operation is not interrupted. A bank of observers provides residuals for fault detection and replacement signals for the reconfiguration. Control is temporarily switched from closed loop into open-loop to decouple the drive from faulty sensor readings. During a short period of open-loop operation, the fault is isolated using parity equations. Replacement signals from observers are used to reconfigure the drive and reenter closed-loop control. There are no large transients in the current. Measurement results and stability analysis show good results.

Current Sensor Fault Detection, Isolation, and Reconfiguration for Doubly Fed Induction Generators

Fault tolerance is gaining growing interest to increase the reliability and availability of distributed energy sources. Current sensor fault detection, isolation, and reconfiguration are presented for a voltage-oriented controlled doubly fed induction generator, which is mainly used in wind turbines. The focus of this analysis is on the isolation of the faulty sensor and the actual reconfiguration. During a short period of open-loop operation, the fault is isolated by looking at residuals calculated from observed and measured signals. Then, replacement signals from observers are used to reconfigure the drive and reenter closed-loop control. Laboratory measurement results are included to prove that the proposed concept leads to good results.

Current Sensor Fault Detection by Bilinear Observer for a Doubly Fed Induction Generator

A bilinear observer is presented to detect current sensor faults in a doubly fed induction generator. The state space model is derived from the voltage equations of stator and rotor. The bilinear behaviour of induction machines and its influence on the state space model is discussed. An open loop model suffers from model and parameter uncertainty. A closed loop observer is shown to greatly reduce these effects. Observer pole placement is mentioned. Feeding back rotor currents only, the observer becomes independent from stator current measurements. The resulting free residuals can be used for sensor fault detection. The observer is validated by laboratory experiment

Current Sensor Fault Detection and Reconfiguration for a Doubly Fed Induction Generator

Sensor fault detection and reconfiguration of the control loops of a doubly-fed induction generator is presented in this paper. The stator and rotor currents are measured as well as observed. During fault free operation, the measured signals are used for the voltage oriented control. In case of a current sensor fault, the faulty measurement is identified and the control is reconfigured using the observer output. Laboratory measurements prove this concept.

Voltage Sensor Fault Detection and Reconfiguration for a Doubly Fed Induction Generator

Fault detection and reconfiguration of the control loops of a Doubly-Fed Induction Generator are described in this paper. The stator voltage is measured as well as observed. During fault free operation, the measured signal is used for the field oriented control. In case of a voltage sensor fault, the faulty measurement is identified and the control is reconfigured using the observer output. Operation without measuring the stator voltage is possible. Laboratory measurements prove this concept.

Current Sensor Fault Detection, Identification, and Reconfiguration for Doubly Fed Induction Generators

This work presents current sensor fault detection, identification and reconfiguration for a voltage oriented controlled doubly fed induction generator. The focus of this analysis is on the identification of the faulty sensor, and the actual reconfiguration. It is proposed to temporary switch from closed loop into open loop control to decouple the drive from faulty sensor readings. During a short period of open loop operation, the fault is identified. Then replacement signals from observers are used to reconfigure the drive and re-enter closed loop control. Measurement results are included to prove that the proposed concept leads to good results.

Model-based fault detection of gain and offset faults in Doubly Fed Induction Generators

Fault tolerance is gaining interest as a means to increase the reliability and availability of distributed energy systems. In this paper, a voltage-oriented Doubly-Fed Induction Generator is examined. This work applies formerly developped model based sensor fault tolerant control methods to gain and offset faults. For that reason, saturation is modelled. Fault detection is triggered by residuals. Fault isolation determines the faulty sensor. Replacement signals from observers are used to reconfigure the drive and re-enter closed loop control. While model based FDI for these faults is generally possible, it requires the fault to be large enough to be detectable.

Grid connection of multi-Megawatt clean Wave energy power plant under weak grid condition

The WaveDragon is a 7 megawatt wave energy converter, that is currently developed for clean offshore energy production. The system will be floating several kilometres off the Pembrokeshire coast, Wales, UK, and transfers energy using a submarine power cable. Key interest is the control of the power take-off system at disturbed voltages and under weak-grid condition due to long submarine cable and remote, rural location. A multi-reference frame controller is implemented for harmonics compensation. A parameter estimator is implemented to estimate the grid impedance. Simulation and measurements results from a laboratory test setup that illustrate properties of developed method are shown.

Advanced sensor fault detection and control reconfiguration of wind power plants using doubly fed induction generators

Fault tolerance is gaining growing interest to increase reliability and availability of distributed energy systems where, a model based current and voltage sensor fault detection, isolation and reconfiguration is presented for a voltage oriented controlled double Fed induction generator, which is mainly used in wind turbines. A bank of observers is used to provide residuals for the fault detection and replacement signals for the reconfiguration of the chosen observers and the fault detection via parity emulation and isolation logic are presented in detail during a short period of open loop operation, the fault is isolated when replacement signals from observers are used to reconfigure the drive and re-enter closed loop control. The concept is analised. There are no transients in the current. The installed Crowbar is not triggered. Measurement results are included to prove that the proposed concept leads to good results.

Online parameter identification methods for doubly fed induction generators

Two different online parameter identification methods for doubly fed induction generators (DFIG) are investigated in this paper. A model reference adaptive system (MRAS) and a new approach for estimation of the inductances are introduced. Lyapunov stability theory is applied to the adaptive law of the MRAS method. This method requires a test signal which excites all systems eigenvalues. Therefore stator and rotor have to be excited by a test signal. However, not all eigenvalues are excited if the stator of the DFIG is directly connected to the grid. In contrast to common methods the new approach presented here needs no excitation signal. This method estimates the inductances by a set of equations which are based on the description of the DFIG neglecting the stator and rotor resistances. It yields to good results for the mutual inductance and the stator leakage inductance also for the grid connected stator. Both methods are analysed by simulations. The new approach for estimation of the inductances is also tested by measurements at a laboratory setup and yields to good results of the stator leakage and the mutual inductances.