Most of Hydro-Québec's generators are nearing the end of their useful life and their ageing is accelerated by phenomena other than those related to their intrinsic mode of operation, such as internal defects. As part of its 2035 vision, which consists of reinventing equipment maintenance to ensure that it is used to its full potential, the company is therefore required to develop non-intrusive diagnostic tools that have no impact on energy production. In addition, the use of alternator fleets at their full potential requires the development of advanced tools allowing the evaluation of their behaviour and the prediction of their performance in the presence of these internal defects. In this context, this research project aims at developing simulation models of large hydroelectric generators that will serve as major inputs for the development of their numerical twins and the evaluation of their behaviour and performance. Non-intrusive diagnostic and prognostic tools will be developed based on the search for fault signatures in the leakage flow and vibro-acoustic measurements using advanced signal processing and artificial intelligence methods. The targeted developments could also be extended to diagnose other electrical equipment such as circuit breakers and transformers. The technology that will be developed can be transferred to other application areas such as transport electrification, marine propulsion, renewable energy and smart grids.