OATAO is an open access repository that collects the work of Toulouse researchers and makes it freely available over the web where possible. This is an author-deposited version published in : http://oatao.univ-toulouse.fr/ Eprints ID : 18858The contribution was presented at ICAART 2017 :http://www.icaart.org/?y=2017 Sensors and actuators are progressively invading our everyday life, as well as industrial processes. They form complex and pervasive systems usually called "ambient systems" or "cyber-physical systems". These systems are supposed to efficiently perform various and dynamic tasks in an ever-changing environment. They need to be able to learn and to self-adapt throughout their life, because designers cannot specify a priori all the interactions and situations they will face. These are strong requirements that push the need for lifelong machine learning, where devices can learn models and behaviors and transfer them to perform other tasks. This article presents a multi-agent approach for lifelong machine learning.

Although wave impact has been extensively studied in laboratories, field studies are comparatively rare. However, as real wave impacts are influenced by numerous environmental factors, complementing physical studies with in-situ data is necessary to better understand the processes at stake and provide reliable tools for coastal engineers. One of the main reasons for the lack of field data is the extreme conditions usually met on site. Nowadays, technology allows to set up stations able to resist those conditions and record data over long periods. In this context, the so-called Artha breakwater, in the French Basque coast, was equipped with an in-situ laboratory to record wave impact pressures. This station enables to collect long term wave impact pressure data therefore covering any weather conditions. In the present paper, the use of computer engineering based methods to process the large amount of wave impact data is described. It involves signal pre-processing, impact automatic segmentation, automatic computation of impact parameters, and artificial intelligence to classify the impacts. Impact automatic segmentation allows to have a big database of impacts available. This database has been used to make a first classification of the strongest impacts. The classification was performed thanks to the parameters automatically computed for each impact. As preliminary results for the classification, several wave impact pressure classes have been established. The approach is encouraging since the obtained results can be compared with the existing laboratory classification. However, the results can still be improved by computing other impact parameters and considering all impacts.

No abstract

No abstract