Abstract: Piecewise linear representation algorithms of time series are usually based on a single heuristic rules，difficult to apply in different characteristics of time series data.Drawing on the idea of edge operator to extract time series edge points，propose a novel piecewise linear representation algorithm that based on temporal edge operator.The experiments on public datasets from several different fields show that compared with two main piecewise linear representation algorithms，the proposed algorithm appears better fitting error and high stability which can be applied to different characteristics of time series data.