The Short-Term Prediction ofWind Power Based on the Convolutional Graph Attention Deep Neural Network

The fluctuation of wind power affects the operating safety and power consumption of the electric power grid and restricts the grid connection of wind power on a large scale.Therefore,wind power forecasting plays a key role in improving the safety and economic benefits of the power grid.This paper proposes a wind power predicting method based on a convolutional graph attention deep neural network with multi-wind farm data.Based on the graph attention network and attention mechanism,the method extracts spatial-temporal characteristics from the data of multiple wind farms.Then,combined with a deep neural network,a convolutional graph attention deep neural network model is constructed.Finally,the model is trained with the quantile regression loss function to achieve the wind power deterministic and probabilistic prediction based on multi-wind farm spatial-temporal data.A wind power dataset in the U.S.is taken as an example to demonstrate the efficacy of the proposed model.Compared with the selected baseline methods,the proposed model achieves the best prediction performance.The point prediction errors(i.e.,root mean square error(RMSE)and normalized mean absolute percentage error(NMAPE))are 0.304 MW and 1.177%,respectively.And the comprehensive performance of probabilistic prediction(i.e.,con-tinuously ranked probability score(CRPS))is 0.580.Thus,the significance of multi-wind farm data and spatial-temporal feature extraction module is self-evident.

Filling Pattern of Volcanostratigraphy of Cenozoic Volcanic Rocks in the Changbaishan Area and Possible Future Eruptions

The Cenozoic volcanostratigraphy in the Changbaishan area had complex building processes.Twenty-two eruption periods have been determined from the Wangtian＇e, Touxi, and Changbaishan volcanoes. The complex volcanostratigraphy of the Changbaishan area can be divided into four types of filling patterns from bottom to top. They are lava flows filling in valleys（LFFV）, lava flows filling in platform（LFFP）, lava flows formed the cone（LFFC）, and pyroclastic Flow filling in crater or valleys（PFFC/V）. LFFV has been divided into four layers and terminates as a lateral overlap. The topography of LFFV, which is controlled by the landform, is lens shaped with a wide flat top and narrow bottom.LFFP has been divided into three layers and terminates as a lateral downlap. The topography of LFFP is sheet and tabular shaped with a narrow top and wide bottom. It has large width to thickness ratio. It was built by multiple eruptive centers distributed along the fissure. The topography of LFFC, which is located above the LFFP, has a hummocky shape with a narrow sloping top and a wide flat bottom. It terminates as a later downlap or backstepping. It has large width to thickness ratio. It was built by a single eruptive center. The topography of PFFC/V, which located above the LFFC, LFFP, or valley, has the shape of fan and terminates as a lateral downlap or overlap. It has a small width to thickness ratio and was built by a single eruptive center. The filling pattern is controlled by temperature, SiO_2 content,volatile content, magma volume, and the paleolandform. In the short term, the eruptive production of the Changbaishan area is comenditic ash or pumice of a Plinian type eruption. The eruptive volume in future should be smaller than that of the Baguamiao period, and the filling pattern should be PFFC/V,which may cause huge damage to adjacent areas.

Prediction of Formation Quality of Inconel 625 Clads Using Support Vector Regression

The process parameters of pulsed tungsten inert gas(PTIG)have a significant infuence on the forma-tion quality,mechanical properties and corrosion resistance of the weld overlay.The PTIG was utilized to deposit Inconel 625 clads with various combinations of the process parameters,which were determined by the central composite design(CCD)method.Based on the experimental results,the relationship between process parameters of PTIG and formation quality of the Inconel 625 clads was established using support vector regression(SVR)with different kernel functions,including polynomial kernel function,radial basis function(RBF)kernel function,and sigmoid kernel function.The results indicate that the kernel functions have a great influence on the prediction of height,width and dilution.The models with RBF kernel function feature the best goodness of fitting and the most accurate against the other SVR models for estimating the height and the dilution.However,the model with polynomial kernel function is superior to the other SVR models for predicting the width.Meanwhile,the prediction performance of the SVR models was compared with the general regression analysis.The results demonstrate that the optimized SVR model is much better than the general regression model in the prediction performance.