Assessment of ship carbon emission reduction technology based on analytical hierarchy and machine learning models

The issue of ship carbon emissions is becoming increasingly prominent. Researchers have proposed various measures to reduce carbon emissions, achieving certain levels of reduction. However, each technique also has its own operational drawbacks. Currently, there is a lack of comprehensive evaluation methods to specifically compare the merits and drawbacks of different technologies. Therefore, this paper proposes the use of Random Forest and Logistic Regression models for prediction of carbon emission reduction technologies. Analytic Hierarchy Process (AHP) to quantitatively evaluate and analyze five typical ship carbon emission reduction technologies. Firstly, the AHP logic is applied to establish a comprehensive evaluation index system for ship carbon emission reduction technologies, encompassing three criteria layers of economy, environment and technology and eleven scheme layers. Secondly, by constructing a judgment matrix, the weight coefficients for each scheme layer are derived, and the overall weights for each criterion are obtained. Finally, three experts were invited to rate the carbon emission reduction technologies, resulting in a ranking of their merits and drawbacks as follows: ammonia fuel technology > carbon capture technology > wind-assisted propulsion technology > air layer drag reduction technology > hydrogen fuel technology. Random Forest model achieves better prediction accuracy compared to Logistic Regression model. The AHP quantitative research method applied in this paper can provide a valuable decision-making support for ship operators.