Data-Driven Anomaly Detection for UAV Sensor Data Based on Deep Learning Prediction Model

Anomaly detection is an active area of research with numerous methods and applications. This survey reviews the state-of-the-art of data-driven anomaly detection techniques and their application to the aviation domain. After a brief introduction to the main traditional data-driven methods for anomaly detection, we review the recent advances in the area of neural networks, deep learning and temporal-logic based learning. In particular, we cover unsupervised techniques applicable to time series data because of their relevance to the aviation domain, where the lack of labeled data is the most usual case, and the nature of flight trajectories and sensor data is sequential, or temporal. The advantages and disadvantages of each method are presented in terms of computational efficiency and detection efficacy. The second part of the survey explores the application of anomaly detection techniques to aviation and their contributions to the improvement of the safety and performance of flight operations and aviation systems. As far as we know, some of the presented methods have not yet found an application in the aviation domain. We review applications ranging from the identification of significant operational events in air traffic operations to the prediction of potential aviation system failures for predictive maintenance.

Download Full-text

Data-Driven Anomaly Detection Based on Multi-Sensor Data Fusion

10.1109/smartnets50376.2021.9555430 ◽

2021 ◽

Author(s):

Di Wang ◽

Ahmad Al-Rubaie ◽

Sandra Stincic ◽

John Davies ◽

Alia Aljasmi

Keyword(s):

Data Fusion ◽

Anomaly Detection ◽

Data Driven ◽

Sensor Data ◽

Sensor Data Fusion ◽

Multi Sensor Data Fusion

Download Full-text

Anomaly Detection and Diagnosis In Manufacturing Systems

Annual Conference of the PHM Society ◽

10.36001/phmconf.2019.v11i1.815 ◽

2019 ◽

Vol 11 (1) ◽

Author(s):

Kalyani Zope ◽

Kuldeep Singh ◽

Sri Harsha Nistala ◽

Arghya Basak ◽

Pradeep Rathore ◽

...

Keyword(s):

Machine Learning ◽

Deep Learning ◽

Anomaly Detection ◽

Manufacturing Systems ◽

Predictive Maintenance ◽

Sensor Data ◽

Statistical Techniques ◽

Industrial Systems ◽

Learning Techniques ◽

Detection And Diagnosis

Multivariate sensor data collected from manufacturing and process industries represents actual operational behavior and can be used for predictive maintenance of the plants. Anomaly detection and diagnosis, that forms an integral part of predictive maintenance, in industrial systems is however challenging due to their complex behavior, interactions among sensors, corrective actions of control systems and variability in anomalous behavior. While several statistical techniques for anomaly detection have been in use for a long time, these are not particularly suited for temporal (or contextual) anomalies that are characteristic of multivariate time series sensor data. On the other hand, several machine learning and deep learning techniques for anomaly detection gained significant interest in the recent years. Further, anomaly diagnosis that involves localization of the faults did not receive much attention. In this work, we compare the anomaly detection and diagnosis capabilities, in semi-supervised mode, of several statistical, machine learning and deep learning techniques on two systems viz. the interacting quadruple tank system and the continuous stirred tank reactor (CSTR) system both of which are representative of the complexity of large industrial systems. The techniques studied include principal component analysis (PCA), Mahalanobis distance (MD), one-class support vector machine (OCSVM), isolation forest, elliptic envelope, dense auto-encoder and long short term memory auto-encoder (LSTM AE). The study revealed that MD and LSTM-AE have the highest anomaly detection capability, followed closely by PCA and OCSVM. The above techniques also exhibited good diagnosis capability. The study indicates that statistical techniques in spite of their simplicity could be as powerful as machine learning and deep learning techniques, and may be considered for anomaly detection and diagnosis in manufacturing systems.

Download Full-text

Recent Advances in Anomaly Detection Methods applied to Aviation

10.20944/preprints201909.0326.v1 ◽

2019 ◽

Author(s):

Luis Basora ◽

Xavier Olive ◽

Thomas Dubot

Keyword(s):

Anomaly Detection ◽

Time Series Data ◽

Data Driven ◽

Sensor Data ◽

Detection Methods ◽

Series Data ◽

Detection Techniques ◽

Advantages And Disadvantages ◽

Recent Advances ◽

Flight Operations

Anomaly detection is an active area of research with numerous methods and applications. This survey reviews the state-of-the-art of data-driven anomaly detection techniques and their application to the the aviation domain. After a brief introduction to the main traditional data-driven methods for anomaly detection, we review the recent advances in the area of neural networks, deep learning and temporal-logic based learning. We cover especially unsupervised techniques applicable to time series data because of their relevance to the aviation domain, where the lack of labeled data is the most usual case, and the nature of flight trajectories and sensor data is sequential, or temporal. The advantages and disadvantages of each method are presented in terms of computational efficiency and detection efficacy. The second part of the survey explores the application of anomaly detection techniques to aviation and their contributions to the improvement of the safety and performance of flight operations and aviation systems. As far as we know, some of the presented methods have not yet found an application in the aviation domain. We review applications ranging from the identification of significant operational events in air traffic operations to the prediction of potential aviation system failures for predictive maintenance.

Download Full-text

Anomaly Detection of 2.4L Diesel Engine Using One-Class SVM with Variational Autoencoder

Annual Conference of the PHM Society ◽

10.36001/phmconf.2019.v11i1.804 ◽

2019 ◽

Vol 11 (1) ◽

Author(s):

Gyebong Jang ◽

Sung-bae Cho

Keyword(s):

Deep Learning ◽

Diesel Engine ◽

Anomaly Detection ◽

Latent Variables ◽

Expert Knowledge ◽

Early Stage ◽

Sensor Data ◽

Robust Solution ◽

Abnormal State ◽

Variational Autoencoder

Despite the intensive research, the study on preventing the breakdown of the construction machine is still at its early stage, so we need to develop an autonomous and robust solution that minimizes equipment downtime and ensures the rigidity of equipment through predictive diagnostics. In particular, engine failure is critical to cause the entire system to stop, so that it is important to determine and predict the symptoms before the failure. However, at present, it is at a level to set specific indicators based on domain knowledge in order to judge the failure. This paper proposes an anomaly detection model for a 2.4L diesel engine, and verify the model using two main faults. The proposed method extracts 130 feature parameters based on autoencoder, which is a deep learning method, and distinguishes between normal and abnormal states by one-class SVM (OCSVM). Autoencoder automatically extracts useful features from multiple sensors on an excavator engine. The data from the engine can represent robust features by using features learned in latent variables using variational autoencoder to extract optimal features. In addition, OCSVM can detect abnormal state and then distinguish between two fault and unknown factors. The experimental results show the accuracy of about 73%, and the false alarm related to the reliability of this abnormality diagnosis model can be minimized to about 17%. Finally, to solve the problem of reliability and analysis of the model itself due to the problem of blackbox, which is a disadvantage of the deep learning model, the LIME analysis method is applied to list the sensor data that affected the determination of the abnormal state. Experts can easily make professional judgments about abnormal conditions and build a model in which known data about faults and symptoms are continuously increasing. The proposed method could improve the accuracy of the model by adding expert knowledge to data-based model.

Download Full-text

A Time-Distributed Spatiotemporal Feature Learning Method for Machine Health Monitoring with Multi-Sensor Time Series

Sensors ◽

10.3390/s18092932 ◽

2018 ◽

Vol 18 (9) ◽

pp. 2932 ◽

Cited By ~ 23

Author(s):

Huihui Qiao ◽

Taiyong Wang ◽

Peng Wang ◽

Shibin Qiao ◽

Lan Zhang

Keyword(s):

Time Series ◽

Deep Learning ◽

Health Monitoring ◽

Data Driven ◽

Sensor Data ◽

Series Data ◽

Spatiotemporal Features ◽

Daunting Task ◽

Machine Health Monitoring ◽

Machine Health

Data-driven methods with multi-sensor time series data are the most promising approaches for monitoring machine health. Extracting fault-sensitive features from multi-sensor time series is a daunting task for both traditional data-driven methods and current deep learning models. A novel hybrid end-to-end deep learning framework named Time-distributed ConvLSTM model (TDConvLSTM) is proposed in the paper for machine health monitoring, which works directly on raw multi-sensor time series. In TDConvLSTM, the normalized multi-sensor data is first segmented into a collection of subsequences by a sliding window along the temporal dimension. Time-distributed local feature extractors are simultaneously applied to each subsequence to extract local spatiotemporal features. Then a holistic ConvLSTM layer is designed to extract holistic spatiotemporal features between subsequences. At last, a fully-connected layer and a supervised learning layer are stacked on the top of the model to obtain the target. TDConvLSTM can extract spatiotemporal features on different time scales without any handcrafted feature engineering. The proposed model can achieve better performance in both time series classification tasks and regression prediction tasks than some state-of-the-art models, which has been verified in the gearbox fault diagnosis experiment and the tool wear prediction experiment.

Download Full-text