NEURAL NETWORK MODELING IN SPORT BIOMECHANICS BASED ON THE EXAMPLE OF SHOT-PUT FLIGHT

  • Klaus D. Maier
  • Petra Meier
  • Heiko Wagner
  • Reinhard Blickhan
Keywords: modeling, neural networks, shot put

Abstract

In this study, a modeling method based on Multi-Layer-Perceptron neural networks (MLP) is presented, using the example of shot-put flight. This method can be used for rapid construction of models. For performance of shot-put, a physical model based on the shots’ equations of motion is easily designed. In this way, an analysis of the shot-put projectory can be used to illustrate the effectiveness of the neural network modeling method. Using the physical model, release data has been determined and altered with random errors such as those introduced by video analysis. A technique for optimal learning with the neural networks has been developed. The resulting MLP models the shot-put flight successfully. The difference between the model’s predicted distance and the distance reached by the physical model are within 2.5%. In conclusion, this method allows rapid creation of models to solve biomechanical problems and can serve as a useful tool for coaches and athletes.