Research on online identification method of ship condition based on improved DBN algorithm

The operating condition of the ship is to describe the change of the required power with time during the navigation of the ship. Accurate real-time identification of the characteristics of the current working conditions and timely adjustment of the relevant control parameters of the strategy have a significant positive effect on further improving the efficiency of energy management. This paper proposes an online identification method of ship condition based on improved deep belief network (DBN). In the traditional DBN algorithm, ship load power data is often considered static, but in fact they are dynamic. Therefore, the differential evolution algorithm is introduced to capture the dynamics of data and apply it to DBN. By constructing the condition identification model based on the improved DBN algorithm, the analysis and judgment of the characteristic parameters of the ship ’s current working condition are carried out to determine the real-time working condition type of the ship. This paper uses real ship load data and compares it with methods based on traditional machine learning models. The experimental results show that it is superior to other algorithms in identification accuracy and operation efficiency. Moreover, it can better capture the dynamics of ship load power data and achieve higher identification rate on different data sets. In summary, the method proposed in this paper is an effective and highly stable ship load power condition identification method, which can be used to optimize the ship energy management strategy and improve the ship operation stability and fuel economy.