Application of reinforcement learning model based on random forest optimization in cost prediction of green power transmission and transformation projects

An optimized reinforcement learning (RL) prediction algorithm based on random forest is proposed to address the issue of low accuracy in cost data prediction for current green power transmission and transformation projects. By analyzing the factors that influence the costs of these projects, they are categorized according to their impact level, filtered using the random forest algorithm, and the selected feature vectors of the influential factors are utilized as training and testing data for the prediction model. Additionally, the cost prediction problem for green power projects is reframed as a transmission line planning issue, and reinforcement learning is employed to optimize the parameters of the ant colony algorithm to construct a cost prediction model for these projects. After experimental comparisons, based on the results from two different grid sizes, the proposed scheme reduced the cost of green power projects by 2.98% and 3.79%, respectively.