This work presents the results of the DETECTA project, which addresses industrial research activities for the generation of predictive knowledge aimed at detecting anomalies in machining-based manufacturing systems. It addresses different technological challenges to simultaneously improve the availability of machinery and the protection against cyberthreats of industrial systems, with the collaboration of knowledge centers and experts in industrial processes. Through the use of innovative technologies such as the digital twin and artificial intelligence, it implements process characterization methodologies and anomaly detection in a non-intrusive way without limiting the productivity of the industrial plant according to the maintenance and remote access needs. The research has been supported by a general evaluation of connected environments in small and medium-sized enterprises to identify if the benefits of digitization outweigh the risks that cannot be eliminated. The results obtained, through a process of supervision by process experts and machine learning, have made it possible to discriminate anomalies between purely technical events and events related to cyber incidents or cyber attacks.