| Maritime transport is one of the most dominant modes of transport and is the backbone of international trade and global supply chains.With the globalization of economy and trade,the transportation industry is becoming more and more dependent on maritime transport,and the ship transportation industry is growing at an accelerated rate,which makes the number of ships grow.However,as the maritime transport industry is mainly powered by heavy fuel oil,it emits a large amount of greenhouse gases and pollutants,and the resulting environmental pollution cannot be ignored.Marine companies and ship owners aim to comply with international environmental regulations and national dual carbon strategy development,focusing on effective reduction of energy consumption and carbon emissions of ships while ensuring commercial profitability of ship operation.Therefore,it is a challenging scientific challenge to improve the energy efficiency of ships.As ship energy consumption is related to basic information of ships,transportation activities and fuel consumption,the key to improve ship energy efficiency is to build an accurate model to predict the fuel consumption rate of ships under different conditions.The main research of this paper is as follows:(1)Data collection related to ship fuel consumption: In the data collection stage,data related to ship fuel consumption are collected through various types of sensors installed on the ship,mainly including speed,heading,draft,wind direction,main engine fuel consumption,etc.After that,the data distribution of various types of sensors was processed for outliers,and the correlation analysis between oil consumption variables and oil consumption-related variables was carried out.According to the a priori knowledge,it is known that the relationship between fuel consumption and each variable is mostly nonlinear.Curve fitting is performed for fuel consumption and variables,and the function relationship with the highest degree of fitting is used to convert the variables into a suitable data distribution.(2)Ship fuel consumption prediction modeling: Firstly,the prediction model takes into account factors such as ship sailing speed,ship weight,sea state and weather conditions,which have non-linear and uncertain effects on ship fuel consumption under different circumstances.Then,the prediction model is proposed based on the nature of the LASSO regression model with sparse variables and the Ridge regression model with overfitting prevention,and the two are combined to construct elastic network regression model.Finally,it is verified that the elastic network regression model not only has the effect of sparse variables and prevents overfitting,but also can effectively solve the problems of multicollinearity and high model complexity caused by high-dimensional variables.It also outperforms the LASSO regression model and Ridge regression model in terms of model prediction accuracy.Compared with the neural network models LSTM and BPNN,the prediction accuracy of the elastic network regression model is higher and more interpretable.(3)Ship fuel consumption optimization analysis: Based on the prediction model proposed in stage(2),the ship speed optimization model is established under the condition of ensuring the arrival at the destination port within the expected arrival time,and the optimization purpose is achieved by constraining the total fuel consumption to be minimized during the ship’s voyage.The solution process uses a mixed integer programming method,which can be solved by Gurobi.In the computational experiments,data from two trajectories with a total of eight voyages were randomly selected for optimization simulation verification.The experimental results show that the optimized speed is analyzed to provide speed decision support for ship voyage planning,and the proposed model can reduce the ship fuel consumption by 0.94%-8.31%.In this paper,a research method of ship fuel consumption prediction and optimization based on elastic network regression model is proposed.The elastic network regression model improves the accuracy of fuel consumption prediction and is more interpretable,which can better describe the relationship between fuel consumption and influencing factors and can provide decision support for crew members during navigation.Speed is the main factor affecting fuel consumption,and minimizing fuel consumption by solving the optimal speed with an optimization model is the easiest and least costly way to optimize energy consumption.The research results prove that the optimization model can not only provide decision support for ship companies to save fuel cost and make sailing plans,but also meet the requirements of increasingly strict regulations to limit greenhouse gas emissions,which has the environmental significance of energy saving and emission reduction. |
