The cyclotron C70XP of the Interest Public Group ARRONAX is regularly producing radio-isotopes for medical and research purposes. To support these productions an internal data network based on EPICS has been deployed, extending the collection of data on the accelerator components and, beam and technical diagnostics. With the accumulation of the new data, a study program is being addressed focusing on the application of data mining and Machine Learning (ML). ML Algorithm, e.g. clusterisation such as density-based spatial clustering of applications with noise or isolation forest, are used to explore the capacity to highlight anomalies for long runs with two extreme temporal cases. First explored approaches and results are presented in this paper as well as the robustness of the algorithms, which are investigated using dedicated methods (indices or iterations).