Neural network based models of javelin flight: prediction of flight distances and optimal release parameters

A model has been developed to predict the flight of javelins using a multi-layer-perceptron neural network. The input parameters to the model are three release angles and the velocity at release, while the output is the distance reached. The inputs and outputs were recorded and analysed for 98 throws. The neural network model was found to predict actual flights of javelins to within 5%, with a mean difference between the model and real throws of 2.5%. The model was used to generate maps of distances reached for different combinations of release parameters. It was found that the most important parameter was the release velocity and that a moderate side angle of attack should be used to attain the longest throws to compensate for rotation of the javelin on release. For release velocities up to 27–28 m s–1 javelins should have an angle of attack about 1–3° larger than the angle of release. For higher velocities this is reversed. In conclusion, the model can be used as a coaching aid to optimise athletes’ throws during training and competition.