Evaluation of fractal dimension of soft terrain surface. (April 2017)

• Record Type: Journal Article
• Authors: Jiang, Chunxia; Lu, Zhixiong; Zhou, Jin; Memon, Muhammad Sohail
• Is Part Of: Journal of terramechanics. Volume 70(2017:Apr.)
• Page Start: 27
• Page End: 34
• Publication Date: 2017-04
• Subjects: Soft terrain surface -- Root mean square method -- Fractal theory -- W–M function; Trafficability -- Periodicals; Praticabilité (Routes) -- Périodiques; Trafficability; Periodicals; 629.222
• DOI: 10.1016/j.jterra.2017.01.003
• Languages: English
• ISSNs: 0022-4898

Abstract

Highlights: Structure function method and rescaled (R/S) analysis method are not suitable on soft terrain because of the precision. The power spectral density method is not robust. Root mean square method is the best method compared with other four methods. The fractal dimension of a plowed surface is smaller than that of a sowed surface. The plowed surface has a more complex internal structure than the sowed surface. Abstract: By using a laser profiler, the roughness of sowed and plowed surfaces was obtained. Through evaluation of the precision of fractal dimensions based on the Weierstrass–Mandelbort (W–M) function, we found the rescaled (R/S) analysis method and the structure spectral method were not suitable for the calculation of the fractal dimension on a soft terrain surface. Therefore, the fractal dimension, non-scale range and correlation coefficient for each kind of terrain were analyzed using the following fractal computational methods – (i) variate-difference method, (ii) power spectral density method, and (iii) root mean square method. The results showed that: fractal dimension of plowed terrain was large with small fluctuations, while its internal structure was complex. The power spectral density method was not robust enough to compute the fractal dimension of a soft terrain surface. The fractal dimension computed using the root mean square method was found to be more accurate for the soft terrain surface. Moreover, the correlation coefficient of linear regression when using the root mean square method was good and the range of non-scale variation was small.