SHORT-TERM POWER PREDICTION FOR DISTRIBUTED PV CLUSTERS BASED ON PCA-SHAPEDTW-QWGRU
A cluster power prediction model based on principal component analysis,shape dynamic time warping and quantum weighted gated recurrent unit(PCA-ShapeDTW-QWGRU)is established for distributed PV short-term power prediction.To address the issue of insufficiently fine cluster division and the difficulty of capturing the information contained in PV plant data,a cluster division method based on PCA with ordering points to identify the clustering structure(PCA-OPTICS)is proposed.Furthermore,to address the issue of low similarity between the selected representative power plants and the clusters,a selection method of representative power plants based on ShapeDTW is proposed,and ShapeDTW measures the similarity of the clusters.ShapeDTW quantifies the similarity distance,identifies the minimum value as the representative power station,and employs the QWGRU model optimised via the root-mean-square propagation(RMSprop-QWGRU)method,to in order to address the discrepancy in the transformation coefficient conversion of the representative power station with the cluster power,real-time transformation coefficients are employed for the prediction calculation of the cluster power value of the representative power station.The proposed method is employed to predict the cluster power value of the representative power station.The experimental results demonstrate that the proposed method can effectively enhance the precision of PV cluster power prediction.
PV power predictionsubgroup divisionprincipal component analysisdynamic time warpingquantum weighted gated recurrent unit
NETL
NSTL
万方数据
