基于小波包分频的自适应模糊神经网络风电功率平抑策略

Adaptive Fuzzy Neural Network Wind Power Leveling Strategy Based on Wavelet Packet Division

风电场频率波动导致风电发出功率不能及时准确跟踪电网调度指令,给电网调度带来很大困难,采用自适应模糊神经网络的小波包分频优化控制策略是一种新的策略,该策略能在风电跟踪调度指令的频率波动范围内优化小波包分频参数,得到优化后的混合储能充放电功率的初始功率指标值,将混合储能荷电状态的平均值和风电场出力与电网调度指令的平均值的偏差作为两个输入,再将风电历史数据训练后的自适应小波包分频模糊神经网络加入控制系统。采用大量风电历史数据在MATLAB/simulink中仿真,结果表明该策略减小了混合储能充放电时对电网频率波动的影响,充放电时混合储能荷电状态波动小,风电跟踪电网调度的效果更优。

The frequency fluctuation of wind farm results in that the output power of wind farms cannot track the dispatching command of power grid timely and accurately,which brings great difficulties to power grid dispatching.The wavelet packet frequency division optimization control strategy using adaptive fuzzy neural network is a new strategy,which can optimize the wavelet packet frequency division parameters within the frequency fluctuation range of wind power tracking dispatching command,and obtain the initial power index value of the optimized hybrid energy storage charging and discharging power.The average value of mixed energy storage state of charge and the deviation between wind farm output and grid dispatching command average value are taken as two inputs,and then the adaptive wavelet packet frequency division fuzzy neural network trained by the wind power historical data is added to the control system.A large amount of wind power historical data is used to simulate in MATLAB/simulink.The results show that this strategy reduces the impact of mixed energy storage charging and discharging on the grid frequency fluctuation,the mixed energy storage charging state fluctuation is small during charging and discharging,and the wind power tracking grid dispatching effect is better.