Simulation Analysis Of Sliding Mode Observer For Fault Detection Of Doubly-fed Wind Turbines

The operating environment of the doubly-fed wind power generation system is changeable,and the status tracking under complex conditions and the fault detection of early faults are of great significance to reduce the maintenance of wind turbines and increase the life of wind turbines.In this paper,three common faults of doubly-fed wind turbine are modeled dynamically,and two sliding mode observers are designed for state tracking and fault detection.The main content of this paper is divided into the following aspects:1.Three common faults in DFIG systems are modeled.Based on the multi-loop theory,the dq model of the doubly-fed wind turbine with any single-phase inter-turn short-circuit fault is deduced.The model can simulate the short-circuit fault between any phase and turns of the generator stator,and can freely adjust the fault damage degree of the stator winding.The current sensor offset fault and grid voltage sag fault are introduced,and their mathematical models are given.The simulation of these three common faults is realized on the MATLAB/SUMULINK simulation platform.2.A switchable approach rate-based sliding mode observer method is proposed for current state tracking and related fault detection of doubly-fed wind turbines.The sliding mode observer based on the switchable approach rate has a switchable term determined by the deviation and the system running time,and it can automatically adjust the gain of the sliding mode control law to achieve the optimal observation effect.Then,the doubly-fed wind power generation system under different conditions is simulated on the MATLAB/SIMULINK platform,and the simulation analysis of the proposed sliding mode observer method is carried out.Simulation experiments show that,in state tracking,the proposed sliding mode observer based on the switchable approach rate can accurately track the system current even under variable wind speed and resistance interference conditions,and the performance tracking accuracy is better than that of the traditional observer.High reliability and stronger robustness;in fault detection,the residual error between the estimated value of the sliding mode observer based on the switchable approach rate and the actual measured value can effectively detect generator stator inter-turn short-circuit faults,current sensor faults and power grid faults Voltage drop failure.3.An adaptive sliding mode observer method based on RBF neural network is proposed.The traditional observer method needs to know the exact model parameter information,when some structural parameter information is unknown,the method will fail.Therefore,a fault detection method of adaptive sliding mode observer based on RBF neural network is designed for the special case of partial parameter loss of the doubly-fed wind turbine system.The fault detection method uses the radial basis neural network to fit the unknown parameter function part of the system model,and then realizes the fault detection of the doubly-fed wind turbine through the residual between the estimated value of the sliding mode observer and the actual measured value.Simulation experiments demonstrate the feasibility of this method in fault detection when part of the system information is lost.