Forecast Of Athletic Achievements Based On Artificial Neural Networks

Introduction
Artificial Neural Network-ANN is a network information system which is build on the basis of brain feedback
information, this system has high nonlinear disposal ability. The article adopts the BP method of artificial neural
network to forecast athletic achievements.

Methods
BP method of artificial neural network, mathematical method

Results
1. Forecast of Athletic Achievements
To illustrate this method, we build a model by the means of numerical value test. We take the national backstroke record
(table 1) from1991to 1997 as learning stylebooks, to forecast later year’s achievement. In this model, we forecast the
achievement of later l year (l=1) according to the achievement of former m year (m=3). According to the model build
before we obtain the result using（table 2）：

Discussion/ Conclusions
year 1990 1991 1992 1993（1994、1995） 1996 1997
serial 1 2 3 4 5 6
yi 2¢05¢¢07 2¢03¢¢952¢03¢¢442¢02¢¢07 2¢01¢¢301¢58¢¢72
Residual test Correlation test Total test
Maximal relative error r P C precision level
0.000842% 0.777778 1 0.000228797 GOOD
Forecast of athletic achievements is actually a nonlinear mapping, BP networks is a nonlinear mapping in which output
varies with input. In this paper, we build a BP model to forecast athletic achievements and obtain higher precision level.
This model could help to study future athletic achievements and assist making developing goal and training decision for
coaches and athletes.

References
[1]Pingfan Yan, Changshui Zhang. Artificial Neural Network and simulate calculation.. Beijing: Tsinghua university
press, 2000
[2]LimingZhang. Model of Artificial Neural Network and its application. Shanghai: Fudan university press,1992
[3]Wenlong Xiao, Jianliang Li. A accelerative convergence method of BP calculation. Enginery transaction. Vol.
23(2002), Suppl., pp113-115

NOTA: O texto com a iconografia está no anexo.

Arquivo