The invention discloses a photovoltaic output prediction method based on correlation. The method comprises the following steps: obtaining photovoltaic output historical data of a reference photovoltaic power station and a target photovoltaic power station in the same region; respectively carrying out non-shielding normalization processing on the acquired historical data of the two power stations,and respectively solving respective non-shielding day j moment output and non-shielding coefficient of each day j moment; calculating a sunny coefficient Z and a mutation coefficient C of each day inthe historical data, performing normalization processing, and performing clustering on the sample days to classify weather; performing function fitting on all sample unshielded coefficients, belongingto certain weather, of the reference photovoltaic power station and the target photovoltaic power station to obtain a cumulative distribution function of the two power stations in various weather; establishing correlation for the cumulative distribution function by using a Copula function; and selecting a spatial correlation model under corresponding weather, solving a non-shielding coefficient of the target photovoltaic power station, and then solving a power prediction result of the target photovoltaic power station. The method is low in cost, small in required data amount and high in prediction precision.