Carbon steel corrosion rate prediction method

The invention discloses a carbon steel corrosion rate prediction method. The method comprises the following steps: constructing a model initial training and test sample according to given historical accumulated natural environment corrosion data; adjusting and optimizing two parameters, namely the number of base classifiers and the maximum tree depth in the random forest by using a genetic algorithm, so as to model and store an optimal random forest prediction model; constructing training of a BP neural network model by using a corrosion prediction result of the initial data sample and selection of important features by using a random forest, and testing the sample; initializing a BP neural network model; using given training and testing samples to optimize, train and test the BP neural network to obtain a modeling optimal neural network prediction model; and inputting the prediction data into the integrated random forest and BP neural network prediction model to obtain a carbon steel corrosion prediction res