Optimization of training load and recovery strategies for track and field sprinters by genetic algorithm

Accurately grasping the training load of athletes is a prerequisite for developing scientific and reasonable training and competition programs. This paper realizes the coupling of the genetic algorithm and BP neural network to design the training load prediction method for athletes based on the IAGABP algorithm. In the genetic algorithm part, N chromosomes are randomly generated using real number coding to form the initial population of the algorithm, and then the genetic operation is continuously performed to improve the overall fitness of the population until the evolution of the population reaches the specified number of generations when the algorithm is terminated. In the BP neural network part, firstly, the structure of the network and other parameters should be determined, and then the optimal individuals obtained in the genetic algorithm part are disassembled into a set of BP neural networks. The connection weights and thresholds are used as the initial weights and thresholds of the BP neural network, and the weights and thresholds are continuously adjusted by error back propagation until the network output error reaches the termination condition and the final network model is obtained. In the dataset test, the mean square error average of this paper’s model is 0.00122875, smaller than that of the GA-BP neural network model, which proves the superior performance of the coupling algorithm. The similarity between the actual and predicted values is 99.15% in the real-world application for athletes. The method of this paper can provide scientific and accurate training load prediction for track and field sprinters, provide a reliable reference for training and competition planning arrangements, and recover from the development of a reasonable plan and standardization of standardized technology.