Abstract: As surface roughness has high randomness and disorder, how to effectively simulate its statistical characteristics becomes the primary prerequisite for the prediction and optimization of interface performance at micro-scale. In order to solve the problem that existing methods cannot satisfy efficiency, accuracy and stability simultaneously, a new reconstruction method is put forward. The proposed method takes sampled data mapping relationship between peak and valley distributions as characteristic carrier and can preserve roughness topography parameters, statistical parameters and asperity distribution. The performance of the proposed method is discussed and is compared with those of existing methods for ideal Gaussian and measured non Gaussian machined surfaces. The results show that the proposed method optimizes the peak-valley mapping process and overcomes the limitations on autocorrelation function form, correlation length and Johnson transformation. In consequence, the proposed method can simulate different types of surfaces and has significant advantages over other methods in efficiency, accuracy and stability.

Key words: roughness, rough surface, surface topography, reconstruction, modeling