Multivariate time series forecasting requires modeling both hierarchical temporal dynamics and complex inter-variable dependencies, a dual requirement that often degrades predictive performance and incurs high computational costs in standard Transformer architectures. Unlike current channel-independent models that ignore vital cross-variable synergies, or dense-attention frameworks that suffer from quadratic computational noise, our approach extracts structurally sparse dependencies. To address these specific limitations, this study introduces the FCA-Transformer. The proposed framework integrates a Feature Pyramid Network (FPN) to isolate macroscopic trends from high-frequency localized fluctuations via hierarchical downsampling. Concurrently, a structured Transformer-based Cross-Attention (TCA) mechanism employs Dimensional Segmentation with Weighting (DSW) and a Two-Stage Attention (TSA) layer to map topological variable interactions, effectively extracting robust cross-variable pathways and mitigating distributional noise. Extensive empirical evaluations across three real-world multivariate benchmarks (ETTh1, Electricity, and Exchange Rate) demonstrate that the FCA-Transformer achieves an average reduction of up to 4.39% in MSE and 5.11% in MAE compared to leading baselines. These findings indicate that the proposed architecture successfully reconciles multi-scale feature extraction with lightweight dependency modeling, enhancing structural generalization and providing a scalable framework for real-time temporal analysis in complex industrial environments.
This paper proposes a feature scheduling aggregation clustering model based on Convolutional Neural Networks (CNN), aiming to enhance the performance of clustering tasks using deep learning techniques. The model integrates the automatic feature extraction capability of CNN, the adaptive adjustment of feature scheduling, and the multi-band information integration of feature aggregation, thereby capturing the complex structures and subtle differences in data. Experimental results demonstrate the superiority of the model compared to existing algorithms and show good generalization ability. The model can effectively improve the effectiveness of cluster analysis, providing a new deep learning solution for data analysis.
This study presents a convolutional neural network (CNN)-based feature-scheduling aggregation clustering framework designed to address the limitations of conventional clustering algorithms in handling high-dimensional data structures. The proposed framework synergistically combines CNN’s automated feature extraction with adaptive feature-scheduling mechanisms and multi-level feature aggregation, enabling the effective capture of nonlinear data relationships and fine-grained pattern distinctions. The experimental results highlight the framework’s superior performance against existing algorithms, alongside its strong generalization capability. This framework advances cluster analysis methodology by providing an integrated deep learning architecture for complex data scenarios.