Search

In view of the fact that most of the current research on power system chain failure is based on traditional power system, and the hidden fault, one of the important factors leading to chain failure, is ignored. Based on this, a novel cascading fault model of power system was constructed based on stochastic load flow (SLF) and cellular automata (CA) algorithms for hidden faults. Firstly, SLF was used to introduce wind power and photovoltaic units to construct a new power system, and the active initial power flow of each line in the system was calculated. Then, the CA algorithm was used to simplify the topology of the system, and the cell state was judged by the hidden fault probability model of power flow over limit, and the state of the cell and its neighbor was used for fault transmission to simulate the chain fault. Finally, the improved IEEE39-node system was taken as an example to verify the effectiveness of the model. This model can provide a new reference for the study of new cascading fault model and hidden fault of power system. [ABSTRACT FROM AUTHOR]

In view of the fact that most of the single or twice decomposed wind power combination prediction models built by using variational modal decomposition (VMD) directly ignore the rich information contained in the residual term of wind power after VMD decomposition, the accuracy of ultra short-term wind power prediction is limited. A combined prediction model based on the quadratic decomposition NGO-VMD residual term、Kmeans clustering algorithm and LSTM was proposed. Firstly, (northern goshawk optimization, NGO) was used to optimize the parameters of VMD to select the best combination of VMD parameters; Secondly, the NGO-VMD model is used to decompose the VMD residual term twice, and the rich information contained in the VMD residual term is deeply mined; Thirdly, the Kmeans clustering algorithm is used to solve the problem that the number of modal components in VMD decomposition is large and the calculation is cumbersome; Finally, the LSTM model is established to predict each sub modal component and the prediction value of each sub modal component is superimposed to obtain the ultra short-term wind power prediction results. The results show that the quadratic decomposition NGO-VMD residual term、Kmeans clustering algorithm and LSTM combined prediction model can fully mine the important information of VMD residual term and effectively improve the accuracy of ultra short-term wind power prediction. [ABSTRACT FROM AUTHOR]

Copyright of Electric Drive is the property of Electric Drive Editorial Office and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

In order to address the problems of different degrees of influence of key meteorological factors on short-term wind power prediction and the lack of prediction accuracy of a single mode, a short-term wind power prediction model based on neighborhood component analysis (NCA) feature weighting and Stacking ensemble prediction is proposed in this paper. Considering the different degrees of influence of meteorological features on wind power, NCA is employed to weight the meteorological features. The weighted features are used as model inputs to enhance the degree of influence of key features. On this basis, multiple base predictors are established to predict wind power, and a combiner is established to integrate the prediction results to build the Stacking ensemble learning prediction model. The analysis of examples shows that the higher accuracy for short-term wind power prediction is achieved with the Stacking ensemble forecasting model using weighted features as input. [ABSTRACT FROM AUTHOR]

Copyright of Journal of Computer Engineering & Applications is the property of Beijing Journal of Computer Engineering & Applications Journal Co Ltd. and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)