Abstract Grid-connected photovoltaic systems show energy losses due to angular, spectral and low irradiance effects. It would be desirable to know the amount of these losses in real systems. Nowadays, we can find in literature several models, which estimate these losses. However, they are not easy to implement. In this paper, a new method that allows the calculation of angular, spectral and low irradiance losses as a whole is presented. It uses an artificial neural network model in combination with the Osterwald model. Both models are integrated in a single structure. The method is an easy-to-use tool, which only receives two inputs: the global irradiance on the plane of the generator and the cell temperature. At present, the method allows the calculation of the mentioned losses as a whole for systems located at Southern Spain.