Abstract
Fault detection (FD) has been important in Industry 4.0 to reduce the cost of manufacturing in the contemporary industry. FD was widely used to determine the health of the medical equipment. Online FD method has been critical in identifying faults in industrial Internet of Things (IIoT) equipment. Today, deep learning (DL) has been developed, which means that deep understanding-related defect detection approaches are gaining more and more popularity, and the extraction of features plays a vital role in such strategies. This can be attributed to the fact that it is difficult to select reliable sensor data manually as input features because of the abundance of sensors. Thus, the proposed research is devoted to the creation of the Hybrid Glowworm Swarm Optimization with Recurrent Deep Learning technique in the IIoT setting Fault Detection (HGSO-RDLFD). The offered HGSO-RDLFD method is aimed at detecting and categorizing faults, as suggested by audio signals in the IIoT setting. To achieve this, the HGSO-RDLFD method has three stages which include feature extraction, fault detection and hyperparameter tuning. Mel spectrogram technique can be used to extract the important features of the audio signal at the initial stage. Next, the HGSO-RDLFD method refers to the repetitive deep learning model that is used to classify faults in the IIoT setting. Then, the HGSO algorithm is played with in the second stage, to achieve the optimum hyperparameter optimization of HCNN-GRU method, and thus lead to the improvement of classification performance. In order to show the potential results of the HGSO-RDLFD approach, a comprehensive experimental analysis is obtained on fault detection dataset. The accuracy of the proposed system was found to be 99.7% that was 5% and 2.4% higher than classical baselines and advanced CNN-LSTM models respectively at a constant statistical significance. The outcomes of the simulation describe the improved results of the HGSO-RDLFD technique compared to the other new methodologies regarding various measures.