ABSTRACT
Masonry towers are an important architectural heritage, whose conservation and maintenance requires a deep understanding of their structural behaviour. To this end, monitoring the dynamic response to ambient and service loads is a fundamental source of information. By repeating the data acquisition over the time, it is moreover possible to check for variations in the structure's response, whose entity may be correlated to the appearance or growth of a damage (e.g. following some exceptional event as an earthquake or as a consequence of materials and components ageing). The complexity of some existing structures and their interaction with the environment claim for a detailed monitoring plan, to support an evidence-based decision process. If the sensor network acquires data continuously over time, the evolution of the structural behaviour may be tracked continuously as well. This process needs the proper methods and algorithms to manage the large amount of available data and extract actionable information from it. This paper presents a methodology for the automatic structural long-term monitoring, which relies on existing methods from the Machine Learning and Data Mining fields. The results of its application to the real-world case of an ancient masonry bell tower, the Ghirlandina Tower (Modena, Italy) are also discussed.
