The simulation of the operating performance and operating parameters of the excavating and conveying system of a cutter suction dredger has always been a research hotspot and difficulty. The single mechanism model has problems such as weak timeliness and large calculation deviations, while the data-driven model faces defects such as poor generalization ability and difficult interpretation of the model. In response to this problem, this paper combines the advantages of the mechanism model reflecting the physical rules and the flexibility of the data-driven method, establishing a dredging operation model of a cutter suction dredger with strong extrapolation generalization ability and high accuracy to realize the numerical simulation of the excavation system and conveying system of the cutter suction dredger. It first processes and analyzes the historical construction data of the cutter suction dredger, make use of the method of information gain rate to select the control variables, and builds the production prediction model of the excavation system through the machine learning method based on improved KNN. Then the excavation production calculated by the excavation system is used as the input of the transportation system, and the improved Euler method in fluid mechanics is used to calculate the segmentation and density of the pipeline, and the mechanism model of the pipeline transportation system of the cutter suction dredger is constructed. Finally, adjust the control parameters in real-time in the excavation and conveying system model, and observe their influence on the results. The analysis and comparison between the numerical results and the actual ship construction data show that the simulation results are accurate and reliable, and can achieve an accurate simulation of the performance and operating conditions of the excavating and conveying system of the cutter suction dredger.
In the process of dredging rivers and filling embankments, engineering equipment such as dredgers plays a key role (Hu et al., 2011). Among them, the cutter suction dredger adopts the continuous operation mode of digging, transporting and unloading. It uses a rotating reamer to cut the underwater soil layer, and then uses a centrifugal pump to transport the muddy water mixture over a long distance. It is widely used in the construction and maintenance of coastal and inland port waterways, dredging of rivers and lakes, and environmental protection dredging construction (Chen et al., 2018; Wang et al., 2021). During the construction process, the flow velocity, concentration and production of the dredger are related to many factors, such as the speed of the mud pump, the speed of the reamer, the traverse speed of the bridge, the stepping displacement of the trolley, the depth of the bridge, etc (Zhang et al., 2015). The influencing factors are numerous and complex, with large randomness and difficulty in predicting. To this end, Wang et al. (2021) of the National Engineering Research Center of Dredging adopted a data-driven approach and established a construction output model of cutter suction dredger based on PCA and RBF neural network, which solved the black box problem between control variables and instantaneous output in the control system of cutter suction dredger. Similar researches on the modeling of cutter suction dredger based on data-driven method include the prediction model of cutter suction dredger output based on data mining method (Bai et al., 2019), the cutter suction dredger model established by double hidden layer BP neural network (Yang et al., 2016), and the cutter suction dredger model considering the influence of time delay (Chen et al., 2019), etc. However, the above-mentioned models all have defects such as the prediction results are only valid within the scope of the sample data, the generalization of the extrapolation is questionable, and the physical meaning is unclear. While the traditional mathematical and physical models established by the working characteristics of the key equipment of the cutter suction dredger, such as the modeling of the driving system of the cutter suction dredger (Hu et al., 2019) and the dynamic theoretical model of the mud transportation of the long-distance pipeline of the cutter suction dredger (Zhou et al., 2021), are faced with computational bias and computational limitations caused by using empirical formulas to simplify the mechanism model.
