With the increasingly widespread use of often interconnected sensors (accelerometers, RFID chips, GPS, etc.), the volume and complexity of the data produced is growing. However, currently available data analysis algorithms in Machine Learning and Data Mining are poorly suited to processing the large numbers of time series produced by these sensors. Moreover, the various tools that make it possible to mine this data are often unconnected, making it necessary to set up time-consuming processing chains with no real added value for the end user.

In response to this problem, as part of the PIA (Programme d’Investissements d’Avenir) program launched by the French Government and founded by BPIFrance we started the IKATS project (Innovative ToolKit for Analysing Time Series). This initiative, involved CS company (http://www.c-s.fr) the Laboratoire Informatique de Grenoble (LIG : https://www.liglab.fr) and supported by Airbus and EDF R&D.

It aims to provide a ready-to-use toolkit that will give the user all the necessary software for handling, exploring/analyzing and visualizing large numbers of time series within a single framework. Analysis of this data will make it possible to determine essential predictive models, for example in the field of predictive maintenance. There are countless potential applications. For manufacturing industries, where sensors are used to monitor and maintain large-scale operating systems (aeronautics, energy, rail), IKATS will make it possible to model and monitor manufacturing lines and processes. It will also to be possible to use it as part of industrial operating systems based on hi-tech products and requiring high-performance supervision by networks of sensors: in particular, energy optimization. Not to mention areas that include connected and monitored objects: smart buildings, quantified self, telecommunications, the military, and so on. The IKATS project comes in addition to PLM solutions offered by CS, as well as digital simulation and high-performance computing solutions.