Research On Soft Sensor Modeling And Predictive Control For Dredging And Loading Process Of Trailing Hopper Suction Dredger

Trailing suction hopper dredgers are widely used in dredging engineering due to their strong maneuverability and suitability for different working conditions.However,currently,trailing suction hopper dredgers are still manually operated,lacking unified guidance standards,unstable dredging efficiency,and even posing safety risks to dredging operations.Therefore,improving the intelligent level of suction dredgers and improving their dredging operation methods has profound significance.This study focuses on the practical problems encountered during the loading process of trailing suction hopper dredgers,and conducts research on system modeling and control methods for the loading process.The specific content is as follows:(1)Analysis and modeling of the loading and sedimentation process of a trailing suction dredger.A mechanism analysis was conducted for the dredging and loading process.Through the analysis of the mud tank mass balance equation and the loading process state equation,mathematical models such as overflow flow rate,overflow density,and loading yield were constructed.At the same time,a hydrodynamic model for the dredging and loading process was established and compared with the mathematical model through CFD simulation.The results showed that the two models had consistency in terms of dry soil weight,overflow density,and overflow loss,with errors within 5%.The effectiveness and reliability of the dredger sedimentation mathematical model have been verified.(2)Based on the BBPSO-ABC algorithm for state estimation during the loading process.There are some important parameters in the sediment mathematical model of the suction dredger that are difficult to measure in engineering,leading to the lack of use of mathematical models to simulate the loading process and guide actual construction.Therefore,three types of particle swarm optimization algorithms were improved to construct soft sensing models.Through comparative experiments,it was found that the BBPSO-ABC algorithm has better soft sensing effects and higher accuracy.(3)Predictive control of loading yield based on fuzzy Koopman operator.In response to the problem of low construction efficiency caused by the main reliance on manual experience during the loading process of trailing suction hopper dredgers,a study was conducted on the yield prediction and control method for the dredging and loading process.The fuzzy neural network algorithm is used to build the prediction system of the loading output of the trailing suction dredger,and the Koopman operator is combined with the fuzzy neural network model to find a better control strategy.The second strategy can increase the dry soil ton output by 9%when only the height of the overflow tube is changed.It shows that the model prediction controller based on the fuzzy Koopman operator can effectively improve the loading output of the dredger and thus improve the dredging efficiency.