Reliable fault detection systems in industrial processes provide pertinent information for improving the safety and process reliability and reducing manpower costs. Here, we present a flexible and efficient fault detection approach based on the continuous ranked probability score (CRPS) metric to detect faults in multivariate data. This approach merges partial least squares (PLS) models and the CRPS metric to separate normal from abnormal features by simultaneously taking advantage of the feature representation ability of a PLS and the fault detection capacity of a CRPS-based scheme. The proposed approach uses PLS to generate residuals, and then apply the CRPS-based chart to reveal any abnormality. Specifically, two monitoring schemes based on CRPS measure have been introduced in this paper. The first approach uses the Shewhart scheme to evaluate the CRPS of the response variables residuals from the PLS model. The second approach merges the CRPS into the exponentially weighted moving average monitoring chart. We assess the effectiveness of these approaches by using real and simulated distillation column data. We also compare the detection quality of PLS-based CRPS charts to that of PLS-based T2, Q, multivariate cumulative sum, and multivariate exponentially weighted moving average methods. Results show that the capacity of the proposed scheme can reliably detect faults in multivariate processes.

Air quality management and monitoring are vital to maintaining clean air, which is necessary for the health of human, vegetation, and ecosystems. Ozone pollution is one of the main pollutants that negatively affect human health and ecosystems. This paper reports the development of an unsupervised and efficient scheme to detecting anomalies in unlabelled ozone measurements. This scheme combines a Deep Belief Networks (DBN) model and a one-class support vector machine (OCSVM). The DBN model accounts for nonlinear variations in the ground-level ozone concentrations, while OCSVM detects the abnormal ozone measurements. The performance of this approach is evaluated using real data from Is`ere in France. We also compare the detection quality of DBN-based detection schemes to that of deep stacked auto-encoders, Restricted Boltzmann Machinesbased OCSVM and DBN-based clustering procedures (i.e., Kmeans, Birch and Expectation Maximization). The results show that the developed strategy is able to identify anomalies in ozone measurements.

Reliable and timely detection of cyber attacks become indispensable to protect networks and systems. Internet control message protocol (ICMP) flood attacks are still one of the most challenging threats in both IPv4 and IPv6 networks. This paper proposed an approach based on Kullback-Leibler divergence (KLD) to detect ICMP-based Denial Of service (DOS) and Distributed Denial Of Service (DDOS) flooding attacks. This is motivated by the high capacity of KLD to quantitatively discriminate between two distributions. Here, the three-sigma rule is applied to the KLD distances for anomaly detection. We evaluated the effectiveness of this scheme by using the 1999 DARPA Intrusion Detection Evaluation Datasets.

Ozone pollution is one of the most important pollutants that have a negative effect on human health and the ecosystem. An effective statistical methodology to detect abnormal ozone measurements is proposed in this study. We used a Deep Belief Network model to account for nonlinear variation of ground-level ozone concentrations, in combination with a one-class support vector machine, for detecting abnormal ozone measurement. We assessed the efficiency of this methodology by using real data from a network of air quality monitoring systems in Isère, France. Results demonstrated the capability of the proposed strategy to identify abnormalities in ozone measurements.

The efficient operation of wastewater treatment plants (WWTPs) is key to ensuring a sustainable and friendly green environment. Monitoring wastewater processes is helpful not only for evaluating the process operating conditions but also for inspecting product quality. This paper presents a flexible and efficient fault detection approach based on unsupervised deep learning to monitor the operating conditions of WWTPs. Specifically, this approach integrates a deep belief networks (DBN) model and a one-class support vector machine (OCSVM) to separate normal from abnormal features by simultaneously taking advantage of the feature-extraction capability of DBNs and the superior predicting capacity of OCSVM. Here, the DBN model, which is a powerful tool with greedy learning features, accounts for the nonlinear aspects of WWTPs, while OCSVM is used to reliably detect the faults. The developed DBN-OCSVM approach is tested through a practical application on data from a decentralized WWTP in Golden, CO, USA. The results from the DBN-OCSVM are compared with two other detectors: DBN-based K-nearest neighbor and K-means algorithms. The results show the capability of the developed strategy to monitor the WWTP, suggesting that it can raise an early alert to the abnormal conditions.

Accurate detection of changes in land cover leads to better understanding of the dynamics of landscapes. This letter reports the development of a reliable approach to detecting changes in land cover based on remote sensing and radiometric data. This approach integrates the multivariate exponentially weighted moving average (MEWMA) chart with support vector machines (SVMs) for accurate and reliable detection of changes to land cover. Here, we utilize the MEWMA scheme to identify features corresponding to changed regions. Unfortunately, MEWMA schemes cannot discriminate between real changes and false changes. If a change is detected by the MEWMA algorithm, then we execute the SVM algorithm that is based on features corresponding to detected pixels to identify the type of change. We assess the effectiveness of this approach by using the remote-sensing change detection database and the SZTAKI AirChange benchmark data set. Our results show the capacity of our approach to detect changes to land cover.

Reliable detection of faults in PV systems plays an important role in improving their reliability, productivity, and safety. This paper addresses the detection of faults in the direct current (DC) side of photovoltaic (PV) systems using a statistical approach. Specifically, a simulation model that mimics the theoretical performances of the inspected PV system is designed. Residuals, which are the difference between the measured and estimated output data, are used as a fault indicator. Indeed, residuals are used as the input for the Multivariate CUmulative SUM (MCUSUM) algorithm to detect potential faults. We evaluated the proposed method by using data from an actual 20 MWp grid-connected PV system located in the province of Adrar, Algeria.

Obstacle detection is an essential element for the development of intelligent transportation systems so that accidents can be avoided. In this study, we propose a stereovisionbased method for detecting obstacles in urban environment. The proposed method uses a deep stacked auto-encoders (DSA) model that combines the greedy learning features with the dimensionality reduction capacity and employs an unsupervised k-nearest neighbors algorithm (KNN) to accurately and reliably detect the presence of obstacles. We consider obstacle detection as an anomaly detection problem. We evaluated the proposed method by using practical data from three publicly available datasets, the Malaga stereovision urban dataset (MSVUD), the Daimler urban segmentation dataset (DUSD), and Bahnhof dataset. Also, we compared the efficiency of DSA-KNN approach to the deep belief network (DBN)-based clustering schemes. Results show that the DSA-KNN is suitable to visually monitor urban scenes.

Principal components analysis (PCA) has been intensively studied and used in monitoring industrial systems. However, data generated from chemical processes are usually correlated in time due to process dynamics, which makes the fault detection based on PCA approach a challenging task. Accounting for the dynamic nature of data can also reflect the performance of the designed fault detection approaches. In PCA-based methods, this dynamic characteristic of the data can be accounted for by using dynamic PCA (DPCA), in which lagged variables are used in the PCA model to capture the time evolution of the process. This paper presents a new approach that combines the DPCA to account for autocorrelation in data and generalized likelihood ratio (GLR) test to detect faults. A DPCA model is applied to perform dimension reduction while appropriately considering the temporal relationships in the data. Specifically, the proposed approach uses the DPCA to generate residuals, and then apply GLR test to reveal any abnormality. The performances of the proposed method are evaluated through a continuous stirred tank heater system.

Efficient management of patient demands in emergency departments (EDs) has recently received increasing attention by most healthcare administrations. Forecasting ED demands greatly helps ED's managers to make suitable decisions by optimally allocating the available limited resources to efficiently handle patient attendances. Furthermore, it permits pre-emptive action(s) to mitigate and/or prevent overcrowding situations and to enhance the quality of care. In this work, we present a statistical approach based on a vector autoregressive moving average (VARMA) model for a short term forecasting of daily attendances at an ED. The VARMA model has been validated using an experimental data from the paediatric emergency department (PED) at Lille regional hospital centre, France. The results obtained indicate the effectiveness of the proposed approach in forecasting patient demands.