关键词:
长短期记忆网络
极限学习机
风电机组
超短期
功率预测
摘要:
风速的不稳定性和不规律性使得风电机组的发电量呈现较大的波动性,从而可能对电网造成扰动,增加电网的调度难度。针对这些不利影响,提出了一种基于长短期记忆网络(LSTM)和极限学习机(ELM)的风电机组超短期功率预测模型。通过提前预测机组功率输出,缓解风电并网所带来的波动性问题。首先利用风电机组风速历史序列数据建立LSTM模型预测下一个调度周期(15 min)的风速;然后利用叶轮、主轴、齿轮箱、发电机等设备测点历史数据和预测风速建立超短期功率预测ELM模型。结合国内某风电场实例数据分析,预测结果和误差分析表明所提方法在超短期功率预测上的有效性,对于提高电网稳定性和可靠性具有重要意义。The instability and irregularity of wind speed make the power generation of wind turbines show large volatility, which may cause disturbances to the power grid and increase the difficulty of grid scheduling. Aiming at these unfavorable effects, an ultra-short-term power prediction model for wind turbines based on long-short-term memory network (LSTM) and extreme learning machine (ELM) is proposed. By predicting the power output of the turbine in advance, the volatility problem caused by the grid integration of wind power is mitigated. Firstly, the LSTM model is established to predict the wind speed in the next dispatch cycle (15 min) by using the historical sequence data of wind speed of wind turbine;then the ELM model is established to predict the ultra-short-term power by using the historical data of measurement points of impeller, main shaft, gear box, generator and other equipment, as well as the predicted wind speed. Combined with the data analysis of a wind farm in China, the prediction results and error analysis show the effectiveness of the proposed method in ultra-short-term power prediction, which is of great significance for improving the stability and reliability of the power grid.