Sports training is an important part of competitive sports activities. Its direct purpose is to improve the competitive ability of athletes, and then through participating in sports competitions, the competitive ability that has been obtained will be transformed into sports results. In the current sports training function monitoring, most of them rely on traditional experience and models, lacking scientific evaluation and diagnostic methods and standards. With the rapid development of modern science and technology, this paper analyzes the athletes’ sports state based on the ANN (artificial neural network) algorithm. The experimental data shows that the sports state monitoring system established in this paper tests the sports ability by adjusting the sports load before the competition. It is found that the load of runners and swimmers fluctuates greatly before the competition, and the load of bicycles fluctuates relatively smoothly, With the increase of the number of athletes in the test, the proportion of the three types of athletes completing the same intensity load, and the level of blood lactic acid has been improved, indicating that the aerobic metabolism ability is improved. The system designed in this paper increases the speed, breadth and depth of sports training data and information collection, and makes timely feedback through comprehensive and systematic analysis of data and information, thus playing an important role in guiding coaches’ scientific decision-making in sports training.