Drilling scheme analysis and output prediction of instrumented wheel-set based on Finite Element Method and LSTM

Abstract

Based on the Finite Element Method and wheel-rail force measurement theory, the effects of different drilling schemes and parameters on the output of the measuring electrical bridge are analyzed, and the optimal parameters are determined. The finite element model of a metro instrumented wheel-set is established by software ABAQUS. The effects of drilling radius, hole size and wheel-rail contact point on the strain of measuring points are calculated and analyzed, and the measuring bridge outputs of two different drilling schemes are compared. The output of the wheel-rail forces is predicted based on LSTM long short memory network. The results show that: (1) the optimal drilling position is on the middle surface of the wheel, (2) The drilling scheme of eight-hole is better than that of four-hole, (3) LSTM neural network can realize decoupling of bridge signal and high precision prediction of wheel-rail forces.