The wheel polygonisation is the harmonic wave along the wheel circumference, which has been widely reported in high-speed railway vehicles in China in recent years. The polygonization mechanism has not been fully understood yet, and this work discusses one possible mechanism through several experiments. Firstly, a field test for the CRH3 high-speed train with polygonal wheels is carried out. It is found that the vibration in vertical takes major responsibility, and there are several resonance bands for the train-track system. Another field test shows that the rail is sensitive to the polygonisation characteristic frequency, but the wheel axle is not. Then the rail local vibration modes are analyzed by the operational deflection shape (ODS) test in the laboratory. A loaded bogie is fixed on the tested track to simulate the real working conditions. The result shows that the frequency of one local mode is 591 Hz, which is a typical wheel polygonisation frequency. Besides, the wheel-rail contact point is not on the vibration node of this local mode. Thus, the resonance of this local rail mode can be one possible mechanism for the high-order wheel polygonisation.@en

Motivation: Microarray gene expression data are increasingly employed to identify sets of marker genes that accurately predict disease development and outcome in cancer. Many computational approaches have been proposed to construct such predictors. However, there is, as yet, no objective way to evaluate whether a new approach truly improves on the current state of the art. In addition no `standard¿ computational approach has emerged which enables robust outcome prediction. Results: An important contribution of this work is the description of a principled training and validation protocol, which allows objective evaluation of the complete methodology for constructing a predictor. We review the possible choices of computational approaches, with specific emphasis on predictor choice and reporter selection strategies. Employing this training-validation protocol, we evaluated different reporter selection strategies and predictors on six gene expression datasets of varying degrees of difficulty. We demonstrate that simple reporter selection strategies (forward filtering and shrunken centroids) work surprisingly well and outperform partial least squares in four of the six datasets. Similarly, simple predictors, such as the nearest mean classifier, outperform more complex classifiers. Our training-validation protocol provides a robust methodology to evaluate the performance of new computational approaches and to objectively compare outcome predictions on different datasets.@en