The rapid growth of IoT has increased the demand for large-scale data processing. However, traditional centralized methods struggle with real-time requirements and data security. This paper introduces VCD-TSNet, a novel real-time IoT data processing framework that combines blockchain and edge computing. By integrating deep learning models like VGG, ConvLSTM, and DNN, VCD-TSNet effectively performs spatial feature extraction, temporal modeling, and decision-making, while using blockchain to ensure data integrity and privacy. Experimental results demonstrate that VCD-TSNet outperforms baseline models in classification accuracy, prediction precision, and real-time performance. For instance, on the BoT-IoT dataset, the classification accuracy reaches 97.5%, throughput increases to 920 TPS, and response time stays below 85 ms. This study validates the model’s effectiveness and highlights its potential in large-scale IoT environments, offering efficient, secure solutions for real-time data processing. It also provides insights for future improvements in frameworks that combine edge computing with blockchain.