The invention relates to a parameter estimation and correction method for a power supply transient process, and belongs to the technical field of energy management. Because the precision of the photovoltaic power supply output prediction model is easily influenced by the state of the photovoltaic power supply output prediction model and a plurality of environmental factors, such as sunlight intensity, temperature, humidity and the like, and the estimation of model parameters is easy to fall into local optimum, a traditional prediction model cannot meet the requirements of safe and stable analysis and configuration of a power grid. According to the method, a prediction scheme based on a recursive least square method is introduced, a relation model between the output value of the photovoltaic power supply to be predicted and historical observation values is established, model parameters are obtained by minimizing prediction errors, and recursive updating of the model parameters is achieved on the basis of a data updating strategy of a sliding rectangular window. By the adoption of the method, model parameters can be conveniently estimated, real-time updating of the model parameters is achieved, prediction precision is higher, the method can well adapt to the situation that power parameters change, and model robustness is greatly improved.