Research On Control Strategy For Low-voltage Ride Through Of Double-fed Fan Based On Artificial Intelligence

With the continuous development of grid-connected technologies,new energy sources such as wind energy and solar energy participate in the generation and regulation of power grids in the form of distributed power sources.Among them,wind power generation has been vigorously researched and developed as the most potential form of new energy power generation,and the number and capacity of grid-connected units are also gradually increasing.Therefore,the impact of the operation status of wind turbines on the system cannot be ignored.Double-fed induction generator,as a representative model of variable-speed constant-frequency generator,has been widely used because of its low cost and easy maintenance.However,the special structure of the unit also reflects its vulnerability and is susceptible to grid disturbances Impact.In order to avoid the large number of wind turbines being disconnected from the grid during grid faults,resulting in greater power shortages,today the requirements for the unit’s low voltage ride through(LVRT)operating capacity are also increasing.The following are the main research methods and steps of this article:First understand the operating principle of DFIG,and study the control strategy of the fan under normal operation by establishing a corresponding mathematical model,on the basis of which to analyze the transient characteristics of the unit during power grid failure.In view of different degrees of power grid faults,this paper proposes two low voltage ride-through schemes,an improved control strategy and a series variable braking resistor.Under mild faults,the control loop structure can be improved to eliminate stator-side transient components to achieve LVRT.In case of serious faults,the crowbar protection circuit is replaced by the rotor string resistance,which can effectively suppress the rotor over-voltage and over-current,keep the converter controllable when it is put into use,and set the series connection of different voltage drop points of the grid connection point.The effective resistance value improves the flexibility of the solution.Most of the current low-voltage ride-through schemes are only effective under certain specific faults and have certain limitations.Considering the complex situation of the actual operation of the power grid,fixed control parameters are often difficult to ensure the low voltage ride-through effect under each fault.For this,the interface module provided in PSCAD / EMTDC is used to call the MATLAB program to control under various faults The parameters are optimized.Although the optimizationparameters based on the PSO algorithm can achieve good control results,this method is an offline optimization method,and the optimization time is long,which cannot meet the calculation speed requirements of real-time control,and the fault transient process is not suitable for practical engineering.among.In order to solve the problem of optimization time,it is proposed to build a parameter optimization model with a deep learning network,and realize the real-time optimization control of the low-voltage ride-through scheme through the idea of "offline training,online calculation".On PSCAD,the different faults of the DFIG-containing power grid under different operating modes are simulated,taking into account the multiple fault locations and the transition resistance,and a large number of fault data is obtained.The k-means algorithm is used to cluster faults with similar transient characteristics according to the peak value of the rotor current.Then,an appropriate LVRT strategy is set for each type of fault,and PSCAD and MATLAB are used to jointly optimize the corresponding strategy parameters.Establish a DNN network,divide the collected fault simulation data into a training set and a test set,and use the optimized parameters as training labels to repeatedly train the network.After the training is completed,when the new fault information is input into the network,the selection of the plan and the optimal parameter can be quickly realized,the problem of poor real-time performance of the offline optimization plan is solved,and the execution efficiency of the DFIG control system is improved.