The CNN-BiLSTM Model for Design of Building Intelligent Monitoring Systems Under Artificial Intelligence

Abstract

This study proposes an improved attention-guided convolutional bidirectional long short-term memory (IAC-BiLSTM) model, which integrates a modified convolutional block attention module (CBAM) with a convolutional neural network (CNN) and BiLSTM architectures. Meanwhile, this model is applied to the behavior recognition of building areas. For seven categories of “transferable behaviors” carefully selected from the University of Central Florida 101 (UCF101) and Human Motion Database 51(HMDB51) datasets, these behaviors have direct correspondences with typical monitoring targets in buildings. This enables effective mapping between public datasets and specific application scenarios. In summary, this study verifies the superior performance and broad application potential of the IAC-BiLSTM model in building behavior recognition. Hence, the study provides a solid theoretical and practical foundation for optimizing intelligent building management systems.