Gymnastics has attracted people’s attention in various competitions by virtue of its beautiful movements and difficult technical performances. The research and exploration of neural network theory are to make the process from receiving instructions to completing actions in one go while sending instructions. Compared with the previous competitive gymnastics, it mainly relies on personal understanding and coach’s guidance to get high-scoring movements. Through intelligent calculation, it can decompose and calculate each movement in detail and adjust its strength, so as to further improve its excellent characteristics. We combined artificial neural network and intelligent optimization method to achieve more scientific and reasonable guidance for gymnastics, so that people can get better performance in performance and competition. The experimental summary of this paper is as follows: (1) Under the analysis of sample data 5000, it shows that the right segmentation curve is superior to the left segmentation curve and is relatively stable. (2) The BP neural network is obviously higher than the other two algorithms in the experimental test under different dimensions, and the qualified dimension value is controlled within the standard range. (3) In the experimental diagram of mean square error, the ideal error value is about 0.005, and it is possible to reach this target value only under the condition of perfect performance. (4) Convergence function is to explore the gracefulness of action analysis, and the convergence of its artificial peak group greatly shows the performance of global exploration.

Gymnastics is an increasingly popular sport and an important event in the Olympic Games. However, the number of unavoidable injuries in sports is also increasing, and the treatment after the injury is very important. We reduce the harm caused by the injury through the identification and research of pictures. Image preprocessing and other methods can in-depth learn about gymnastics sports injuries. We identify the injured pictures of athletes to know the injury situation. Through the analysis of the force of the athletes during exercise, they can be better integrated into picture recognition for sports injuries. More appropriate prevention and treatment measures are suggested.