The article proposes a methodological framework to develop a digital twin of a photovoltaic power plant, using electrical and meteorological data. As a case study, the solar plant on Baltra Island in Galapagos is used, where variables such as current, voltage, irradiance and temperature were collected. A neural network LSTM (Long Short-Term Memory) was applied due to its ability to model complex temporal relationships in data series. The methodology includes data collection, synchronisation, cleaning and scaling, time sequence training, LSTM model design and performance evaluation with metrics such as MAE and RMSE. The results show high pre-accuracy in the prediction of generated power. The study highlights the processing of large volumes of sensor data stored in Excel, automatic cleaning, temporal visualisation and predictive modelling. It is concluded that it is feasible to build robust digital twins for photovoltaic systems, although advanced technical skills in AI and data processing are required. This approach offers key benefits for sustainable energy management, maintenance and planning in sensitive environments such as the Galapagos Islands.