Browsing by Keyword "Time series analysis"
Now showing 1 - 3 of 3
Results Per Page
Sort Options
Item A hyper-heuristic inspired approach for automatic failure prediction in the context of industry 4.0(2022-09) Navajas-Guerrero, Adriana; Manjarres, Diana; Portillo, Eva; Landa-Torres, Itziar; IAIn the era of technological advances and Industry 4.0, massive data collection and analysis is a common approach followed by many industries and companies worldwide. One of the most important uses of data mining and Machine Learning techniques is to predict possible breaks or failures in industrial processes or machinery. This research designs and develops a hyper-heuristic inspired methodology to autonomously identify significant parameters of the time series that characterize the behaviour of relevant process variables enabling the prediction of failures. The proposed hyper-heuristic inspired approach is based on the combination of an optimization process performed by a meta-heuristic algorithm (Harmony Search) and feature based statistical methods for anomaly detection. It demonstrates its adaptability to different failure cases without expert domain knowledge and the capability of autonomously identifying most relevant parameters of the time series to detect the abnormal behaviour prior to the final failure. The proposed solution is validated against a real database of a cold stamping process yielding satisfactory results respect to a novel AUC_ROC based metric, named AUC_MOD, and other conventional metrics, i.e., Specificity, Sensitivity and False Positive Rate.Item Intelligent maintenance for industrial processes, a case study on cold stamping(Springer Verlag, 2018) Boto, Fernando; Lizuain, Zigor; Cortadi, Alberto Jimenez; Perez Garcia, Hilde; Alfonso-Cendon, Javier; Sanchez Gonzalez, Lidia; Corchado, Emilio; Quintian, Hector; Tecnalia Research & Innovation; FACTORYThe correct diagnosis of tool breakage is fundamental to improve productivity, minimizing the number of unproductive hours and avoiding expensive repairs. The use of Data Mining techniques provides a significant added value in terms of improvements in the robustness, reliability and flexibility of the monitored systems. In this work, a general view of a diagnosis and prognosis of tool breakage in Industrial Processes is proposed. The important issues identified will be analyzed: filtering, process characterization and data based modeling. A case study has been implemented to carry out the prognosis of tool breakage in the cold stamping process. The results provided are qualitative trends and hypothesis to perform the prognosis. Although a validation in real operation is needed, these results are promising and demonstrate the goodness of using these type of techniques in real processes.Item Robust image classification against adversarial attacks using elastic similarity measures between edge count sequences(2020-08) Oregi, Izaskun; Del Ser, Javier; Pérez, Aritz; Lozano, José A.; Quantum; IADue to their unprecedented capacity to learn patterns from raw data, deep neural networks have become the de facto modeling choice to address complex machine learning tasks. However, recent works have emphasized the vulnerability of deep neural networks when being fed with intelligently manipulated adversarial data instances tailored to confuse the model. In order to overcome this issue, a major effort has been made to find methods capable of making deep learning models robust against adversarial inputs. This work presents a new perspective for improving the robustness of deep neural networks in image classification. In computer vision scenarios, adversarial images are crafted by manipulating legitimate inputs so that the target classifier is eventually fooled, but the manipulation is not visually distinguishable by an external observer. The reason for the imperceptibility of the attack is that the human visual system fails to detect minor variations in color space, but excels at detecting anomalies in geometric shapes. We capitalize on this fact by extracting color gradient features from input images at multiple sensitivity levels to detect possible manipulations. We resort to a deep neural classifier to predict the category of unseen images, whereas a discrimination model analyzes the extracted color gradient features with time series techniques to determine the legitimacy of input images. The performance of our method is assessed over experiments comprising state-of-the-art techniques for crafting adversarial attacks. Results corroborate the increased robustness of the classifier when using our discrimination module, yielding drastically reduced success rates of adversarial attacks that operate on the whole image rather than on localized regions or around the existing shapes of the image. Future research is outlined towards improving the detection accuracy of the proposed method for more general attack strategies.