Research On Design Method Of Combustion System Of Marine Diesel Engine Based On Target Heat Release Rate Coupled With Artificial Intelligence

As pollution and energy scarcity become increasingly critical,the issue of CO₂ emissions has received considerable attention.Emissions of pollutants,fuel consumption and engine combustion are closely related,and good combustion of the engine depends on many parameters,including fuel injection pressure,nozzle diameter,fuel injection angle,piston geometry parameters,inlet parameters,etc.This constitutes a multi-input and multi-output modeling and optimization problem.Currently available optimization methods generally optimize the engine from the perspective of combustion theory or numerical methods,but cannot comprehensively utilize the advantages of both to accelerate optimization speed.Therefore,based on a certain type of medium-speed marine diesel engine,this paper propose a method that combines combustion concepts with artificial intelligence to optimize the engine combustion system（nozzle diameter,injection pressure,spray angle,and combustion chamber structure）,significantly reducing the calculation required for optimization design.The main work of this paper is as follows:A mathematical model for the effective thermal efficiency and the specific heat at constant volume was developed based on the Sabathe-Miller ideal Cycle.Considering the restrictions of detonation pressure and pressure rise rate,the optimal specific heat at constant volume and ideal heat release process for the current compression ratio were determined.Using the double Weber function to fit the heat release rate according to the premixed/diffusion combination ratio of the ideal heat release process,the optimal target heat release rate is calculated based on the characteristic parameters of the heat release rate under the current boundary conditions.The relationship between the target heat release rate and the fuel injection parameters and the combustion chamber shape was established.By combining the analytical formula for the combustible fuel mass in the premixed combustion stage with the fuel mass in the premixed combustion stage with the fuel injection nozzle diameter and injection pressure,and the formula for the spray penetration distance,the nozzle diameter and injection pressure were directly calculated.The optimized parameters were validated by three-dimensional simulations,and the premixed combustion process corresponding to the target heat release could be achieved approximately.Based on theoretical optimization,a numerical optimization method driven by three-dimensional simulations data was used to optimize the combustion chamber shape and spray angle to further match the target heat release rate.First,the engine combustion chamber was modeled by parameterization using CAESES software to obtain four independent variables（bowl angle,bore ratio,bulge width,and spray angle）.Then orthogonal experimental design combined with range analysis method was used to analyze the effects of independent variables on engine fuel consumption and emissions using 49 sets of CFD simulations data,the design range of independent variables was narrowed down,which helps to reduce optimization time.This paper proposes a new type of optimizer based on active learning-memetic algorithm for conducting more 3D simulation data-driven optimization searches within the feasible range of variables.The optimizer uses an optimization result-based and bipartite graph querying strategy to concentrate the computational resources around the global optimum while also taking into account the search for the model’s uncertain space.A total of 368 sets of CFD simulations data were used to optimize the combustion chamber shape and fuel injection angle.Compared with the machine learning model-optimization algorithm trained with the same amount of random data,the new optimizer has better model accuracy and more stable and accurate optimization results.The final combustion system approximated by active learning-memetic algorithm optimization achieves the target combustion process and enhances the utilization of air in the cylinder to improve the combustion efficiency of diesel engines,thereby improving engine efficiency.Under the requirements of emission regulation and energy load,fuel consumption is reduced by 14.3%compared to the original engine.