Wind speed, wind direction, humidity, temperature, altitude, and other factors affect wind power generation, and the uncertainty and instability of the above factors bring challenges to the regulation and control of wind power generation, which requires flexible management and scheduling strategies. Therefore, it is crucial to improve the accuracy of ultra-short-term wind power prediction. To solve this problem, this paper proposes an ultra-short-term wind power prediction method with MIVNDN. Firstly, the Spearman’s and Kendall’s correlation coefficients are integrated to select the appropriate features. Secondly, the multi-strategy dung beetle optimization algorithm (MSDBO) is used to optimize the parameter combinations in the improved complete ensemble empirical mode decomposition with adaptive noise (ICEEMDAN) method, and the optimized decomposition method is used to decompose the historical wind power sequence to obtain a series of intrinsic modal function (IMF) components with different frequency ranges. Then, the high-frequency band IMF components and low-frequency band IMF components are reconstructed using the t-mean test and sample entropy, and the reconstructed high-frequency IMF component is decomposed quadratically using the variational modal decomposition (VMD) to obtain a new set of IMF components. Finally, the Nons-Transformer model is improved by adding dilated causal convolution to its encoder, and the new set of IMF components, as well as the unreconstructed mid-frequency band IMF components and the reconstructed low-frequency IMF, component are used as inputs to the model to obtain the prediction results and perform error analysis. The experimental results show that our proposed model outperforms other single and combined models.
Deep learning is often inadequate for achieving effective dialect recognition in situations where data are limited and model training is complex. Differences between Mandarin and dialects, such as the varied pronunciation variants and distinct linguistic features of dialects, often result in a significant decline in recognition performance. In addition, existing work often overlooks the similarities between Mandarin and its dialects and fails to leverage these connections to enhance recognition accuracy. To address these challenges, we propose the Decoupled Fusion Network (DFNet). This network extracts acoustic private and shared features of different languages through feature decoupling, which enhances adaptation to the uniqueness and similarity of these two speech patterns. In addition, we designed a heterogeneous information-weighted fusion module to effectively combine the decoupled Mandarin and dialect features. This strategy leverages the similarity between Mandarin and its dialects, enabling the sharing of multilingual information, and notably enhance the model’s recognition capabilities on low-resource dialect data. An evaluation of our method on the Henan and Guangdong datasets shows that the DFNet performance has improved by 2.64% and 2.68%, respectively. Additionally, a significant number of ablation comparison experiments demonstrate the effectiveness of the method.