Abstract: A corrosion rate prediction algorithm based on KPCA-GA-BP combination model was proposed for the internal corrosion of shale gas gathering and transportation pipelines. Using the detection results of a shale gas gathering and transportation pipeline as training data, a prediction model was established using backpropagation (BP) neural network. Genetic algorithm (GA) was used to optimize the initial values of neural network weights and thresholds, and kernel principal component analysis (KPCA) was used to reduce the dimensionality of the data. During the model building process, the accuracy of the model's prediction was continuously improved. The constructed model was used to predict and excavate another adjacent pipeline for verification. The results show that when selecting TRAINGDM as the training function, the hidden layer nodes was (8, 1), the genetic algorithm evolution number was 50, the population size was 100, the crossover probability was 0.3, and the mutation probability was 0.2, after using KPCA to reduce the 7 dimensions to 4 dimensions, the minimum mean square error (MSE) of this model was 0.12. When this model was used for predicting adjacent pipelines, the MSE was 0.14. The application of KPCA-GA-BP model to predict the internal corrosion rate of shale gas gathering and transportation pipelines had certain accuracy, and this model could be used to assist in guiding the direct evaluation of on-site corrosion and other related work.