| With the continuous maturation and rapid development of wind power generation technology,the single unit capacity of wind turbines is increasing,and its internal structure is becoming more and more complicated.Variable pitch wind turbines have become the mainstream model of modern large-scale wind turbines with its unique advantages.As one of the core components of wind turbine control system,variable pitch system plays an important role in improving the safety,efficiency and stability of wind turbines.Variable pitch system is a highly nonlinear and strongly coupled system,which integrating mechanical,electrical and hydraulic components,because of the bad operating environment and frequent start and stop of the unit,it causes the frequent failure of the variable pitch system,which results in high maintenance costs.Therefore,it is of great significance to study the fault diagnosis of variable pitch system.However,due to the impact of the actual operating environment,there are many uncertain factors such as random noise,vibration disturbance and unknown input in the the variable pitch system,which brings great difficulty to fault diagnosis.As a nonlinear filtering method,particle filter has been widely researched and applied in the field of fault diagnosis,because of that it is not constrained by the assumption such as the linearization of the model and the noise obeys Gaussian distribution.But the traditional particle filter algorithms exist some problems that particle degeneracy and particle impoverishment,which leads to low accuracy of fault state estimation.Therefore,by mainly analyzing the working principle and fault mechanism of the variable pitch system,this paper conduct a research about the fault state estimation method based on particle filter and its fault diagnosis strategy,and the main research contents are as follows:(1)By analyzing the wind turbine system structure,working principle and the relationship between the main subsystems of the wind turbine.Then,focus on analyzing the structure,working principle and failure mechanism of the variable pitch system.Finally,based on the dynamic model of variable pitch system,a fault diagnosis model was established under the framework of particle filter.(2)In order to solve the problem that the lower estimation precision caused by the particle degeneracy and particle impoverishment of the traditional particle filter in the fault state estimation of the pitch system,in this paper,an adaptive bat algorithm is effectively fused with the particle filter to study it.Firstly,due to the strong optimization characteristic of the bat algorithm,the adaptive bat algorithm is incorporated into the particle sampling process,and the latest measurement value is used to define the particle’s fitness function,which guides the particles to move to the high-likelihood region.Secondly,a dynamic adaptive inertia weight is introduced to optimize the global position updating mechanism of particles,which improves the problem of particle impoverishment and trapping into local extremum,thereby improving the accuracy of state estimation.(3)Due to the influence of uncertain factors such as random noise,coupled interference in the actual operating environment,and exists modeling errors in the actual research of variable pitch system,the fault diagnosis accuracy is not high when used the traditional fixed threshold to make fault decision.In this paper,the improved adaptive threshold is used as a residual discriminant function to study it.By analyzing the residual information of the fault,the adaptive threshold is designed by using the idea of the confidence interval in statistics,and by calculating the recursive estimation formula of the mean and variance of the residual to optimize the adaptive threshold.,so as to improve the accuracy and real-time performance of fault diagnosis. |
