Abstract: The existence of measurement error is inevitable in a variety of measurement systems. Combining with the advantages of the traditional filter methods, the gradient mean prediction filtering algorithm was proposed in view of the random errors existence in slowly changing measurement data. The basic idea of the algorithm and flow were described.A filtering experiment was done using the measurement data of bridge continuous deflection measurement system. The result indicated that the algorithm can effectively remove the interference signal in slowly changing data, and the filter effect is obvious.

Key words: Measurement error, Gradient mean prediction, Filtering, Algorithm