Method for predicting corrosion resistance of ammonia-diesel dual fuel four-stroke internal combustion engine valve materials

Abstract

The application particularly relates to a method for predicting corrosion resistance of a gas valve material of an ammonia-diesel dual-fuel four-stroke internal combustion engine, which comprises the following steps: constructing a gas valve corrosion prediction model based on a deep learning model; determining optimal model parameters of the gas valve corrosion prediction model through a genetic algorithm; performing a gas valve corrosion test, collecting relevant parameters, and determining first environmental parameters affecting a corrosion speed; performing a gas valve corrosion simulation test, collecting relevant data, and determining second environmental parameters affecting the corrosion speed; constructing a test database, and extracting a plurality of sets of training data from the test database; training the gas valve corrosion prediction model with the optimal model parameters through the training data until the model converges; and inputting material properties of a valve to be predicted, target environmental parameters and a corrosion time into the trained gas valve corrosion prediction model, and outputting a corresponding predicted corrosion speed. The application improves the accuracy and effectiveness of the prediction of the corrosion resistance of the gas valve material through the combination of a GA algorithm and a BP neural network.

Images