Abstract: There is a problem that the implicit increment information in data can be eliminated when processing the data of time series in existing dynamic comprehensive evaluation by static dimensionless methods. To solve the problem, three improved methods are suggested, i.e. the standard-series, whole-series and increment-weight methods, with a numerical example given to analyze their characteristics. The results showed that if there is no special subjective preference between transverse and lengthways information, the whole-series method is an explicit and easy one available to extend well to various static dimensionless methods and keep on the data characteristics of original method fully. It is therefore recommended to adopt the whole-series method prior to others. Moreover, the thoughts as above could be applied to dynamic non-linear dimensionless problems.