Acoustic Anomaly Detection Using Convolutional Autoencoders in Industrial Processes

In the industrial plants, detection of abnormal events during the processes is a difficult task for human operators who need to monitor the production. In this work, the main aim is to detect anomalies in the industrial processes by an intelligent audio based solution for the new generation of factories. Therefore, this paper presents a Convolutional Autoencoder (CAE) based end-to-end unsupervised Acoustic Anomaly Detection (AAD) system to be used in the context of industrial plants and processes. In this research, a new industrial acoustic dataset has been created by gathering the audio data obtained from a number of videos of industrial processes, recorded in factories involving industrial tools and processes. Due to the fact that the anomalous events in real life are rather rare and the creation of these events is highly costly, anomaly event sounds are superimposed to regular factory soundscape by using different Signal-to-Noise Ratio (SNR) values. To show the effectiveness of the proposed system, the performances of the feature extraction and the AAD are evaluated. The comparison has been made between CAE, One-Class Support Vector Machine (OCSVM), and a hybrid approach of them (CAE-OCSVM) under various SNRs for different anomaly and process sounds. The results showed that CAE with the end-to-end strategy outperforms OCSVM while the respective results are close to the results of hybrid approach.

[1]  Christopher Leckie,et al.  High-dimensional and large-scale anomaly detection using a linear one-class SVM with deep learning , 2016, Pattern Recognit..

[2]  Rakesh Bobba,et al.  Learning Based Anomaly Detection for Industrial Arm Applications , 2018, CPS-SPC@CCS.

[3]  Justin Salamon,et al.  Deep Convolutional Neural Networks and Data Augmentation for Environmental Sound Classification , 2016, IEEE Signal Processing Letters.

[4]  Daniel A. Hartman.  Real-Time Detection of Processing Flaws During Inertia Friction Welding of Critical Components , 2012 .

[5]  Kuang-Ku Chen,et al.  Integrating support vector machine and genetic algorithm to implement dynamic wafer quality prediction system , 2010, Expert Syst. Appl..

[6]  Francesco Piazza,et al.  An End-To-End Unsupervised Approach Employing Convolutional Neural Network Autoencoders for Human Fall Detection , 2019, Quantifying and Processing Biomedical and Behavioral Signals.

[7]  Erik Marchi,et al.  Deep Recurrent Neural Network-Based Autoencoders for Acoustic Novelty Detection , 2017, Comput. Intell. Neurosci..

[8]  Francesco Piazza,et al.  Acoustic novelty detection with adversarial autoencoders , 2017, 2017 International Joint Conference on Neural Networks (IJCNN).

[9]  Jay Lee,et al.  Assessment of Data Suitability for Machine Prognosis Using Maximum Mean Discrepancy , 2018, IEEE Transactions on Industrial Electronics.

[10]  Javier Del Ser,et al.  Data fusion and machine learning for industrial prognosis: Trends and perspectives towards Industry 4.0 , 2019, Inf. Fusion.

[11]  Erik Marchi,et al.  A novel approach for automatic acoustic novelty detection using a denoising autoencoder with bidirectional LSTM neural networks , 2015, 2015 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP).

[12]  Rene de Jesus Romero-Troncoso,et al.  Vibration-based adaptive novelty detection method for monitoring faults in a kinematic chain , 2016 .

[13]  Daniel P. W. Ellis,et al.  Learning Sound Event Classifiers from Web Audio with Noisy Labels , 2019, ICASSP 2019 - 2019 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP).

[14]  M. Popescu,et al.  Acoustic fall detection using one-class classifiers , 2009, 2009 Annual International Conference of the IEEE Engineering in Medicine and Biology Society.

[15]  Miguel Delgado Prieto,et al.  Enhanced Industrial Machinery Condition Monitoring Methodology Based on Novelty Detection and Multi-Modal Analysis , 2016, IEEE Access.

[16]  W. Youden,et al.  Index for rating diagnostic tests , 1950, Cancer.

[17]  Vincenzo Moscato,et al.  One-Class SVM Based Approach for Detecting Anomalous Audio Events , 2014, 2014 International Conference on Intelligent Networking and Collaborative Systems.

[18]  Karol J. Piczak.  Environmental sound classification with convolutional neural networks , 2015, 2015 IEEE 25th International Workshop on Machine Learning for Signal Processing (MLSP).