The paper presents an approach which, combines the particle swarm algorithm and gray forecast model, can effectively improve the Meter accuracy. The implementation of particle swarm nonhomogeneous grey system, using MATLAB programming, uses a small amount of the instrument known error data as training data to get the fullscale error fitting model. Then the model can be used to amend all values in the range. The system has the advantages of economy, practicality and high accuracy characteristic proved by experiments.