Predictive maintenance attempts to prevent unscheduled downtime by scheduling maintenance before expected failures and/or breakdowns while maximally optimizing uptime. However, this is a non-trivial problem, which requires sufficient data analytics knowledge and labeled data, either to design supervised fault detection models or to evaluate the performance of unsupervised models. While today most companies collect data by adding sensors to their machinery, the majority of this data is unfortunately not labeled. Moreover, labeling requires expert knowledge and is very cumbersome. To solve this mismatch, we present an architecture that guides experts, only requiring them to label a very small subset of the data compared to today’s standard labeling campaigns that are used when designing predictive maintenance solutions. We use auto-encoders to highlight potential anomalies and clustering approaches to group these anomalies into (potential) failure types. The accompanied dashboard then presents the anomalies to domain experts for labeling. In this way, we enable domain experts to enrich routinely collected machine data with business intelligence via a user-friendly hybrid model, combining auto-encoder models with labeling steps and supervised models. Ultimately, the labeled failure data allows for creating better failure prediction models, which in turn enables more effective predictive maintenance. More specifically, our architecture gets rid of cumbersome labeling tasks, allowing companies to make maximum use of their data and expert knowledge to ultimately increase their profit. Using our methodology, we achieve a labeling gain of 90% at best compared to standard labeling tasks.

Detection of defects and identification of symptoms in monitoring industrial systems is a widely studied problem with applications in a wide range of domains. Most of the monitored information extracted from systems corresponds to data series (or time series), where the evolution of values through one or multiple dimensions directly illustrates its health state. Thus, an automatic anomaly detection method in data series becomes crucial. In this article, we propose a novel method based on a convolutional neural network to detect precursors of anomalies in multivariate data series. Our contribution is twofold: We first describe a new convolutional architecture dedicated to multivariate data series classification; We then propose a novel method that returns dCAM, a dimension-wise Class Activation Map specifically designed for multivariate time series that can be used to identify precursors when used for classifying normal and abnormal data series. Experiments with several synthetic datasets demonstrate that dCAM is more accurate than previous classification approaches and a viable solution for discriminant feature discovery and classification explanation in multivariate time series. We then experimentally evaluate our approach on a real and challenging use case dedicated to identifying vibration precursors on pumps in nuclear power plants.

Population-based structural health monitoring (PBSHM) provides a means of accounting for inter-turbine correlations when solving the problem of wind farm anomaly detection. Across a wind farm, where a group of structures (turbines) is placed in close vicinity to each other, the environmental conditions and, thus, structural behavior vary in a spatiotemporal manner. Spatiotemporal trends are often overlooked in the existing data-based wind farm anomaly detection methods, because most current methods are designed for individual structures, that is, detecting anomalous behavior of a turbine based on the past behavior of the same turbine. In contrast, the idea of PBSHM involves sharing data across a population of structures and capturing the interactions between structures. This paper proposes a population-based anomaly detection method, specifically for a localized population of structures, which accounts for the spatiotemporal correlations in structural behavior. A case study from an offshore wind farm is given to demonstrate the potential of the proposed method as a wind farm performance indicator. It is concluded that the method has the potential to indicate operational anomalies caused by a range of factors across a wind farm. The method may also be useful for other tasks such as wind power and turbine load modeling.

The US government invests substantial sums to control the HIV/AIDS epidemic. To monitor progress toward epidemic control, PEPFAR, or the President’s Emergency Plan for AIDS Relief, oversees a data reporting system that includes standard indicators, reporting formats, information systems, and data warehouses. These data, reported quarterly, inform understanding of the global epidemic, resource allocation, and identification of trouble spots. PEPFAR has developed tools to assess the quality of data reported. These tools made important contributions but are limited in the methods used to identify anomalous data points. The most advanced consider univariate probability distributions, whereas correlations between indicators suggest a multivariate approach is better suited. For temporal analysis, the same tool compares values to the averages of preceding periods, though does not consider underlying trends and seasonal factors. To that end, we apply two methods to identify anomalous data points among routinely collected facility-level HIV/AIDS data. One approach is Recommender Systems, an unsupervised machine learning method that captures relationships between users and items. We apply the approach in a novel way by predicting reported values, comparing predicted to reported values, and identifying the greatest deviations. For a temporal perspective, we apply time series models that are flexible to include trend and seasonality. Results of these methods were validated against manual review (95% agreement on non-anomalies, 56% agreement on anomalies for recommender systems; 96% agreement on non-anomalies, 91% agreement on anomalies for time series). This tool will apply greater methodological sophistication to monitoring data quality in an accelerated and standardized manner.

Research among bilinguals suggests a foreign language effect for various tasks requiring a more systematic processing style. For instance, bilinguals seem less prone to heuristic reasoning when solving problem statements in their foreign (FL) as opposed to their native (NL) language. The present study aimed to determine whether such an effect might also be observed in the detection of semantic anomalies. Participants were presented NL and FL questions with and without anomalies while their eye movements were recorded. Overall, they failed to detect the anomaly in more than half of the trials. Furthermore, more illusions occurred for questions presented in the FL, indicating an FL disadvantage. Additionally, eye movement analyses suggested that reading patterns for anomalies are predominantly similar across languages. Our results therefore substantiate theories suggesting that FL use induces cognitive load, causing increased susceptibility to illusions due to partial semantic processing.

Anomaly detection in asset condition data is critical for reliable industrial asset operations. But statistical anomaly classifiers require certain amount of normal operations training data before acceptable accuracy can be achieved. The necessary training data are often not available in the early periods of assets operations. This problem is addressed in this paper using a hierarchical model for the asset fleet that systematically identifies similar assets, and enables collaborative learning within the clusters of similar assets. The general behavior of the similar assets are represented using higher level models, from which the parameters are sampled describing the individual asset operations. Hierarchical models enable the individuals from a population, comprising of statistically coherent subpopulations, to collaboratively learn from one another. Results obtained with the hierarchical model show a marked improvement in anomaly detection for assets having low amount of data, compared to independent modeling or having a model common to the entire fleet.

Anomaly detection can be seen as an unsupervised learning task in which a predictive model created on historical data is used to detect outlying instances in new data. This work addresses possibly promising but relatively uncommon application of anomaly detection to text data. Two English-language and one Polish-language Internet discussion forums devoted to psychoactive substances received from home-grown plants, such as hashish or marijuana, serve as text sources that are both realistic and possibly interesting on their own, due to potential associations with drug-related crime. The utility of two different vector text representations is examined: the simple bag of words representation and a more refined Global Vectors (GloVe) representation, which is an example of the increasingly popular word embedding approach. They are both combined with two unsupervised anomaly detection methods, based on one-class support vector machines (SVM) and based on dissimilarity to k-medoids clusters. The GloVe representation is found definitely more useful for anomaly detection, permitting better detection quality and ameliorating the curse of dimensionality issues with text clustering. The cluster dissimilarity approach combined with this representation outperforms one-class SVM with respect to detection quality and appears a more promising approach to anomaly detection in text data.

The use of flexible and autonomous robotics systems is the solution for the automation task of the production and intra-logistics environments. This dynamic context requires the robot to be aware of its surroundings through the whole task, also after accomplishing the gripping action. We present an anomaly detection approach based on unsupervised learning and reconstruction fidelity of image data. We design our method to enhance the dynamic environment perception of robotics systems and apply it in a palletizing robot, in order to perceive and detect changes to its surrounding and process after the gripping step. Our proposed approach achieves the performance targeted by the considered industrial requirements.

Maritime anomaly detection can improve the situational awareness of vessel traffic supervisors and reduce maritime accidents. In order to better detect anomalous behaviour of a vessel in real time, a method that consists of a Density-Based Spatial Clustering of Applications with Noise (DBSCAN) algorithm and a recurrent neural network is presented. In the method presented, the parameters of the DBSCAN algorithm were determined through statistical analysis, and the results of clustering were taken as the traffic patterns to train a recurrent neural network composed of Long Short-Term Memory (LSTM) units. The neural network was applied as a vessel trajectory predictor to conduct real-time maritime anomaly detection. Based on data from the Chinese Zhoushan Islands, experiments verified the applicability of the proposed method. The results show that the proposed method can detect anomalous behaviours of a vessel regarding speed, course and route quickly.

Erection planning in shipbuilding is a highly complex process. When a process change happens for some reason, it is often difficult to identify how many factors are affected and estimate how sensitive these factors can be. To optimize the planning and replanning of the shipbuilding plan for the best production performance, a data-driven approach for shipbuilding erection planning is proposed, which is composed of an erection plan model, identification of major factors related to the erection plan, and a data-driven algorithm to apply shipbuilding operation data for creating plans and forecasting, for plan adjustment, future availabilities of shipyard resources including machines, equipment, and man power. Through data clustering, the relevant factors are identified as a result of plan change, and critical equipment health management is carried out through data-driven anomaly detection. A case study is implemented, and the result shows that the proposed data-driven method is able to reschedule the shipbuilding plans smoothly.

The problem of analysing a single vessel's behaviour from real but incomplete Automatic Identification System (AIS) data received by satellite has been investigated. The main objective was to infer the route of any single vessel of interest, utilising the dynamic information decoded from AIS messages. A complete process of route inference using position, speed, course over ground and time stamp information is proposed in this paper. Due to the incompleteness of satellite AIS messages, an algorithm incorporating random deviations is also presented to account for the missing sections of obtained vessel routes. Analysis results from a set of real AIS data have demonstrated the applicability of the proposed algorithms in various scenarios.