Adaptive symbol regression method based on multi-task genetic programming algorithm

The invention discloses a self-adaptive symbol regression method based on a multi-task genetic programming algorithm, and the method comprises the following steps: firstly constructing a scalable gene expression coding mode for uniformly coding a plurality of different problems, initializing population individuals according to the coding mode, and carrying out the evaluation on all tasks; then, according to crossover parameter selection, performing type selection crossover operation or guide mutation operation on the population, and in the process, by adopting a self-adaptive control strategy on the crossover parameter, adjusting an evolution behavior so as to improve the search efficiency; next, a connection selection operation is executed to select an individual entering the next generation, and a self-adaptive computing resource redistribution mechanism is constructed to enable a task with relatively backward performance to obtain more computing resources. According to the method, crossover mutation operat