In this article, to meet the requirements of marine engine room simulator on both the simulation speed and simulation accuracy, a mean value engine model (MVEM) for the 7S80ME-C9.2 marine two-stroke diesel engine was developed and validated in the MATLAB/Simulink environment. In consideration of the significant influence of turbocharger compressor on both the engine steady state performance and transient response, a novel compressor model (mass flow rate and isentropic efficiency model) based on a previous study carried out by the first author was proposed with the aim of achieving satisfactory simulation accuracy within the whole engine operating envelope. The predictive and extrapolative capability of the proposed compressor model was validated by carrying out simulation experiments and analyzing the simulation results under steady state condition and during transient process. To make the traditional MVEM capable of predicting in-cylinder pressure trace, the cylinder pressure analytic model proposed by Eriksson and Andersson for the four-stroke SI (spark ignition) engine was adapted to the 7S80ME-C9.2 marine two-stroke diesel engine based on the characteristic of in-cylinder pressure trace of this type of engine and then coupled to the MVEM developed in this paper. Since there is no need to solve any differential equation as it is done in the 0-D model, the advantage of MVEM in running speed is not impaired. For achieving satisfactory simulation accuracy by using the analytic model, the model parameters were calibrated elaborately by using engine measured data and a 0-D model and the relevant tuning procedure was discussed in detail.

In response to the typical fault issues encountered during the operation of marine diesel engines, a fault diagnosis method based on a convolutional neural network (CNN), a temporal convolutional network (TCN), and the attention mechanism (ATTENTION) is proposed, referred to as CNN-TCN–ATTENTION. This method successfully addresses the issue of insufficient feature extraction in previous fault diagnosis algorithms. The CNN is employed to capture the local features of diesel engine faults; the TCN is employed to explore the correlations and temporal dependencies in sequential data, further obtaining global features; and the attention mechanism is introduced to assign different weights to the features, ultimately achieving intelligent fault diagnosis for marine diesel engines. The results of the experiments demonstrate that the CNN-TCN–ATTENTION-based model achieves an accuracy of 100%, showing superior performance compared to the individual CNN, TCN, and CNN-TCN methods. Compared with commonly used algorithms such as Transformer, long short-term memory (LSTM), Gated Recurrent Unit (GRU), and Deep Belief Network (DBN), the proposed method demonstrates significantly higher accuracy. Furthermore, the model maintains an accuracy of over 90% in noise environments such as random noise, Gaussian noise, and salt-and-pepper noise, demonstrating strong diagnostic performance, generalization capability, and noise robustness. This provides a theoretical basis for its practical application in the fault diagnosis of marine diesel engines.

Identifying and analyzing the engine performance and emission characteristics under the condition of performance decay is of significant reference value for fault diagnosis, condition-based maintenance, and health status monitoring. However, there is a lack of relevant research on the currently popular marine large two-stroke dual fuel (DF) engines. To fill the research gap, a detailed zero-/one-dimensional (0D/1D) model of a marine two-stroke DF engine employing the low-pressure gas concept is first established in GT-Power (Version 2020) and validated by comparing the simulation and measured results. Then, three typical types of turbocharger performance decays are defined including turbine efficiency decay, turbine nozzle ring area decay, and turbocharger shaft mechanical efficiency decay. Finally, the three types of decays are introduced to the engine simulation model and parametric runs are performed in both diesel and gas modes to identify and analyze their impacts on the performance and emission characteristics of the investigated marine DF engine. The results reveal that turbocharger performance decay has a significant impact on engine performance parameters, such as brake efficiency, engine speed, boost pressure, etc., as well as CO₂ and NOx emissions, and the specified limit value on certain engine operational parameters will be exceeded when turbocharger performance decays to a certain extent. The changing trend of engine performance and emission parameters as turbocharger performance deteriorates are generally consistent in both operating modes but with significant differences in the extent and magnitude, mainly due to the distinct combustion process (Diesel cycle versus Otto cycle). Furthermore, considering the relative decline in brake efficiency, engine speed drop, and relative increase in CO₂ emission, the investigated engine is less sensitive to the turbocharger performance decay in gas mode. The simulation results also imply that employing a variable geometry turbine (VGT) is capable of improving the brake efficiency of the investigated marine DF engine.

The visual monitoring of ship-engine-room equipment is an essential component of ship-cabin intelligence. In response to issues such as imbalanced quantities of different categories of engine room equipment and severe occlusion, this paper presents improvements to YOLOv8-M. Firstly, the introduction of the SPPFCSPC module enhances the feature extraction capabilities of the backbone extraction network. Subsequently, improvements are implemented in the neck network to create GCFPN, facilitating further feature fusion, and introducing the Dynamic Head module, which fuses the deformable convolution, in the part of the detection head, so as to improve the performance of the network. Finally, the FOCAL EIOU LOSS is introduced, while mitigating the impact of dataset imbalance through class-wise data augmentation. In this paper, the ship cabin equipment dataset and the public dataset MS COCO2017 are evaluated. Compared with YOLOv8-M, the mAP50 of GCD-YOLOv8 is improved by 2.6% and 0.4%, respectively.