Hybrid Deep Networks for Anomaly Detection in Industrial IoT Data

Industrial Internet-of-Things (IIoT) systems generate large volumes of heterogeneous sensor and network data. Ensuring security and operational reliability requires timely detection of anomalies in this data. In this work, we propose a hybrid deep learning approach that combines Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs) to automatically detect anomalies in IIoT data. Our model uses CNN layers to extract local feature patterns from multivariate time-series streams and LSTM-based RNN layers to capture temporal dynamics, followed by a classification layer to identify anomalous events. We validate the approach on an industrial case study (e.g. multi-sensor elevator monitoring) and benchmark datasets, demonstrating high detection accuracy and low false alarm rates. The results indicate that the CNN–RNN hybrid model can effectively capture both spatial and temporal characteristics of IIoT data, outperforming single-network baselines.

کلیدواژه ها:

Industrial IoT ، Anomaly detection ، Deep learning ، Convolutional Neural Network (CNN) ، Recurrent Neural Network (RNN) ، Hybrid model ، Case study