Abstract: In order to solve the problems of difficult parameter determination and low prediction accuracy of submarine pipeline corrosion rate prediction model, a new corrosion rate prediction method based on RF-GOA-RVM was proposed. Random forest (RF) was used to screen the corrosive factors of submarine pipelines, and the main corrosion factors were determined; the grasshopper optimization algorithm (GOA) was used to optimize the correlation vector machine (RVM) parameters to predict the corrosion rate of the pipeline. Simulation experiments show that compared with Particle swarm optimization-correlation vector machine (PSO-RVM) and RVM, RF-GOA-RVM has better model stability and higher prediction accuracy, which can provide decision-making basis for submarine pipeline corrosion failure prediction.

Key words: relevance vector machine, grasshopper optimization algorithm, random forest, submarine pipeline, corrosion rate prediction