Intelligent predictions on frictional properties of non-asbestos brake shoe for mine hoister based on ANN model

Abstract

According to many tribological experiments of non-asbestos brake shoe for mine hoists the authors had investigated before, the original experimental data which contain the influencing rules of braking conditions on frictional properties were obtained. Based on the artificial neural network (ANN) technology and the experimental data swatches, a BP neural network model was established to predict the frictional properties of the brake shoe. Three parameters of braking conditions (braking pressure, sliding velocity and surface temperature) were selected as input vectors. And two parameters of frictional performance (friction coefficient and its stability coefficient) were selected as output vectors. The contrast of prediction values with experimental results shows that the neural network model can predict properly the influencing rules of braking conditions on frictional performance. What is more, the neural network model has quite favorable ability for forecasting the values of both friction coefficient and its stability coefficient. The mean prediction error is less than 5%. Therefore, the neural network model is considered feasible and valuable for predicting of frictional properties for frictional materials.