Stéphane Lecoeuche

Mining frequent episodes aims at recovering sequential patterns from temporal data sequences, which can then be used to predict the occurrence of related events in advance. On the other hand, gradual patterns that capture co-variation of complex attributes in the form "When X increases/decreases, Y increases/decreases" play an important role in many real world applications where huge volumes of complex numerical data must be handled. More recently, they have received attention from the data mining community for exploring temporal data and methods have been defined to automatically extract gradual patterns from temporal data. However, to the best of our knowledge, no method has been proposed to extract gradual patterns that always appear at the identical time intervals in the sequences of temporal data, despite the knowledge that such patterns may bring in certain applications. In this paper, we propose to extract co-variations of periodically repeating attributes from the sequences of temporal data that we call seasonal gradual patterns. We discuss the specific features of these patterns and propose an approach for their extraction by exploiting motif mining algorithms in a sequence, and justify its applicability to the gradual case. Illustrative results obtained from a real world data set are described and show the interest for such patterns.

Mining frequent episodes aims at recovering sequential patterns from temporal data sequences, which can then be used to predict the occurrence of related events in advance. On the other hand, gradual patterns that capture co-variation of complex attributes in the form "When X increases/decreases, Y increases/decreases" play an important role in many real world applications where huge volumes of complex numerical data must be handled. More recently, they have received attention from the data mining community for exploring temporal data and methods have been defined to automatically extract gradual patterns from temporal data. However, to the best of our knowledge, no method has been proposed to extract gradual patterns that always appear at the identical time intervals in the sequences of temporal data, despite the knowledge that such patterns may bring in certain applications. In this paper, we propose to extract co-variations of periodically repeating attributes from the sequences of temporal data that we call seasonal gradual patterns. We discuss the specific features of these patterns and propose an approach for their extraction by exploiting motif mining algorithms in a sequence, and justify its applicability to the gradual case. Illustrative results obtained from a real world data set are described and show the interest for such patterns.

Summary: A switched nonlinear system subject to disturbances is considered in this paper. The switching signal admits an average dwell time and a state feedback control depending on the system operating modes, detected with a maximum time delay, is applied to the system. In this framework, the input‐to‐state stability problem of the closed‐loop system is addressed. Based on some established existence conditions of mode‐dependent Lyapunov‐like functions, the values of the maximum time delay and the average dwell time that allow to achieve the input‐to‐state stability of the closed‐loop system are determined. In order to obtain more tractable results, the existence conditions of the mode‐dependent Lyapunov‐like functions are given in terms of sum‐of‐squares programming in the case of polynomial nonlinearities. In the linear case, they are expressed in terms of linear matrix inequalities and a procedure for the synthesis of the mode‐dependent controller is provided in this situation. The established theoretical results are illustrated through a control problem of a building ventilation system and a switched control problem of a vehicle suspension system.

This paper extends the authors' previous work on a diagnosis algorithm for single open-circuit faults in inverters to the double fault case. The proposed approach is entirely data-driven, using measurements of the stator currents. The first contribution of the present work resides in new choices for the feature variables which allow us to detect double open-circuit faults. These feature data are clustered to different classes corresponding to faulty modes using Auto-adaptive and Dynamical Clustering (AUDyC). These classes are then labelled appropriately which allows us to determine conditions for the fault detection and isolation. Moreover, the values of the algorithm parameters and their influence on the detection time are presented. The proposed improvements are validated on simulation data.