基于PCA-LSTM模型的风电机网相互作用预测
Prediction of Interaction Between Grid and Wind Farms Based on PCA-LSTM Model
-
摘要: 随着风电在电力系统中渗透率不断提高,风电机组接入电网带来的机网相互作用问题已严重影响电网安全和电能质量,对风电机网相互作用进行研究意义重大。在TensorFlow深度学习框架下,提出一种基于长短期记忆(long short-termmemory,LSTM)网络的风电机网相互作用预测模型。首先,通过主成分分析法(principal component analysis,PCA)对多变量时间序列做筛选,降低数据维度。其次,用LSTM网络对选出的风电机网相互作用关联因素序列和风电实际输出序列之间的非线性关系进行建模,并通过实例与其他预测方法对比证明其具有更高的精确度和适用性。最后,对机网相互作用观测对象的预测数据进行Prony分析,通过实测数据验证采用观测对象预测值分析机网相互作用的可行性和有效性。Abstract: In power system with high penetration of wind power, the interaction between gird and wind turbines has a critical impact on power grid and power quality. Study on the interaction between gird and wind turbines is of great significance. A wind turbine-grid interaction prediction model based on long short-term memory（LSTM） network under TensorFlow framework was presented. First, multivariate time series were screened by principal component analysis（PCA） to reduce data dimensions. Second, LSTM network was used to model nonlinear relationship between selected sequence and actual output sequence of wind farm, it is proved that the proposed method has higher accuracy and applicability compared with other prediction methods. Finally, Prony algorithm was used to analyze prediction result of the selected data. According to experiment, the feasibility and effectiveness of the proposed method for analyzing the interaction between grid and wind turbines with forecasting results of selected observation objects are verified.
下载:
