Research On Optimal Power Control Strategy Of Offshore Wind Farm Based On Neural Networks

Due to the complexity of offshore wind farm topology,it is difficult to accurately describe the equivalent model in the traditional power flow calculation method using weighted average parameters.At the same time,due to the rapid development of wind power generation and its increasing proportion in the grid capacity,the power control of wind farm is particularly important for the grid.Therefore,in order to solve the problems which are difficult to describe accurately and lack of effective power distribution in power flow calculation of wind farms and to achieve the purpose of fast and accurate power supply capacity evaluation and optimal power distribution,an equivalent model of dual neural network is constructed by using steady-state mathematical model,neural network and improved particle swarm optimization algorithm.Firstly,the steady-state mathematical model combined with neural network method is adopted to explore the problem that wind farm is difficult to be accurately described in power flow calculation in this paper.Taking the doubly fed induction generator(DFIG)as an example,a simple back-propagation(BP)neural network model is established to be equivalent to a single DFIG.The training data of the neural network is obtained according to the actual DFIG simulation model,in which the wind speed,the stator and rotor voltage are taken as the inputs,and the active and reactive power are taken as the outputs.Furthermore,using the actual 102 MW wind farm with 34 DFIGs,taking the statistical wind speed,equivalent voltage and their data errors as inputs,a BP model based on doubly fed wind turbine is established and trained.The simulation results show that the BP neural network can effectively model the wind farm and quickly evaluate the power supply capacity of the wind farm,so that it can solve the problem because the wind farm is difficult to be simplified in the power flow calculation of the power system,and it has high accuracy.Moreover,aiming at the problem of optimal power distribution of wind farms,the improved particle swarm optimization algorithm combined with neural network method is investigated in this paper.According to the power regulation mode and wind turbine status of DFIG in offshore wind farm,the power adjustable range of each wind turbine is obtained,which is taken as both the optimization variable and constraint condition,and the error between the output of wind farm and grid dispatching instruction is ensured to be within a small enough range,which is also taken as the constraint condition.Taking the minimum line loss of wind farm as the optimization objective,the optimized data of power distribution is obtained by the improved particle swarm optimization algorithm,and the optimized data is used as the learning data of artificial neural network aggregation model.Taking the active and reactive power demand of the grid side at the exit of the wind farm and the wind speed of each wind turbine as the inputs of the artificial neural network,and the active and reactive power reference values of the direct power control of each DFIG as the outputs,a four-layer 36-102-102-68 node artificial neural network aggregation model composed of 34 3MW DFIG generators in an offshore wind farm is established.According to the above method,the problem of fast optimal distribution of wind power is solved.The simulation results show that the proposed method can reduce the line loss in the power adjustable region with good performance and the characteristics of fast response.Thus,the problem of fast and accurate power distribution is solved.Finally,aiming at the problem of building power control system of offshore wind farms,the combination of the above-mentioned equivalent neural model and the power distribution neural network model and forms a top-down wind power system are researched in this paper.By analyzing and comparing the characteristics of the optimized traditional model,the non optimized traditional model,the optimized neural network model and the non optimized neural network model,it is found that the optimized neural network model is the most suitable model for wind farm optimal power control,which solves the problem of offshore wind farm optimization system construction.The model has the characteristics of fast response,small error and excellent performance index.It can accurately and quickly feedback the power supply capacity to the grid dispatching side and complete the instructions issued by the grid dispatching side,and realize the power optimization strategy internally wind farm equivalence and its power control have obvious advantages.