Based on the measured spectra of rail roughness and track structures longitudinal roughness, the rail grinding limit is studied with the help of an established coupled dynamic metro vehicle–track model and a rolling contact fatigue model. The results indicate that metro rail grinding control should be regulated according to corrugation wavelength range and operating speed. Based on the rolling contact fatigue model, longer wavelength of rail corrugation has less influence on the wheel rolling contact fatigue. For the metro lines with a maximum operating speed of 80 km/h, the average levels of rail corrugation in the wavelength ranges of 30–65 mm, 65–125 mm, and 125–250 mm should be less than 5.4, 24.8, and 33.8 dB re 1 μm, respectively; for the ones with the operating speed of 80–120 km/h, the corresponding average corrugation levels in the three wavelength ranges should be less than 4.4, 9.8, and 29.8 dB re 1 μm, respectively.@en

For a heavy-haul locomotive within a wheel repairing period, wheel polygonal wear with different operating mileages is obtained by field testing. The test results show that the maximum radial runout of the wheel can increase to 0.87 mm and accompany with the typical damage of wheel tread shelling. Taking the wheel polygons as input excitation, the locomotive-track coupled dynamic model is established, which is verified by the comparisons of test and calculated wheelset vertical acceleration in time and frequency domains. The variable wheel-rail friction coefficient is introduced so as to consider the dry and wet rail conditions. The wheel-rail dynamic contact characteristics under the traction and dry-wet rail surface conditions are analysed simultaneously. It is found that the wheel polygon deteriorates the locomotive traction performance and induces the obvious wheel-rail slipping with large tangential stress, especially in wet rail condition. In dry condition, the wheel-rail could contact generally in the adhesion state. But the longitudinal creep forces fluctuate locally with some larger amplitudes closed to the adhesion force, which is mainly attributed to the excitation of serious wheel polygon. Comparing with the results of the newly repaired wheel, the maximum wheel-rail vertical force, longitudinal force, normal stress and tangential stress at the end of wheel repairing period can increase by 55 kN, 28 kN, 240 MPa and 470 MPa in sequence. The wheel-rail slipping and high-stress state in traction condition should be the dominant factors contributing to the wheel damage of tread shelling.@en

Cardiac magnetic resonance (CMR) is used extensively in the diagnosis and management of cardiovascular disease. Deep learning methods have proven to deliver segmentation results comparable tohuman experts in CMR imaging, but there have been no convincing results for the problem of end-to-end segmentation and diagnosis from CMR. This is in part due to a lack of sufficiently large datasets required to train robust diagnosis models. In this paper, we proposea learning method to train diagnosis models, where our approach isdesigned to work with relatively small datasets. In particular, the optimisation loss is based on multi-task learning that jointly trains for the tasks of segmentation and diagnosis classification. We hypothesize that segmentation has a regularizing effect on the learning of features relevant for diagnosis. Using the 100 training and 50 testingsamples available from the Automated Cardiac Diagnosis Challenge (ACDC) dataset, which has a balanced distribution of 5 cardiac diagnoses, we observe a reduction of the classification error from 32% to 22%, and a faster convergence compared to a baseline without segmentation. To the best of our knowledge, this is the best diagnosis results from CMR using an end-to-end diagnosis and segmentation learning method.@en