Aiming at several problems existing in the collection of data reading of smart meters, combining dynamic initial value and metabolic modeling ideas, a method based on the improved GM(1,1) model for the second reading and judgment of collection data of smart meters is proposed. Take the data in the original data sequence as the initial value of the GM(1,1) model in turn, infer the initial value corresponding to the smallest residual error, and then obtain the data dimension required to minimize the residual error of the GM(1,1) model, And then use the metabolic modeling idea to establish an improved GM(1,1) model, and apply the model to the second study and judgment of the collective reading data of the smart meter that failed the first billing. The results show that, compared with the traditional least squares method and the traditional GM(1,1) model, the improved GM(1,1) model has better prediction accuracy, and is more suitable for the secondary accounting research and judgment of the collected data of the smart meter.