Study Of Neuro-Optimization Technique For Numerical Simulation Of Casting Process

Filling and solidification of casting is intricate and dynamic process characterized by difficulty of precise control and prediction. This make extension of leading-time and low efficiency. This project try to explore a new way of optimization for numerical simulation of casting process. Through applying artificial intelligence to present numerical simulation method, a procedural optimizing mode is launched. Up to date, expert system is applied extensively in foundry. It can simulate ways of problem solving by means of symbol-based reasoning based on knowledge and experience of casting expert. Due to the fact that expert system depends intensively upon perfection of rules and precision of data, it is unsuitalble to treat such complicated process as numerical simulation of casting process.In this project, based on application of neuro-optimization to numerical simulation of casting process, a number of works are carried out as follows:(1) A mode of foundry engineering design is achieve by means of neuro-optimization. It is applied successfully to design of casting process.(2) Precision and efficiency are two crucial indexes for evaluation of numerical simulation of casting process. The study takes the two indexes as goal of optimization and sets up a systematic method of neuro-optimization. Based on present mode of coupling simulation of casting, feasibility of application of neuro-optimization is discussed. Three ways of optimization is schemed, i.e. CPU time optimizing for finite element modeling, nuero-optimization of casting processparameters. Optimization of free surface algorithm during filling of casting.(3)The study applys the neuro-optimization technique to coupling simulation of casting-ironed camshaft. Resulted from neuro-optimization and FEM respectively, relative error and CPU time in simulation of temperature field of casting are discussed, combined with experimental data. With different pouring schemes of plate casting, neuro-simulation is implemented. It provides a new way of optimal selection to casting schemes. Finally, the algorithm of neuro-optimation is applied to simulation free surface of SOLA-VOF method.The study shows that it is feasible to apply neuro-optimization to numerical simulation of casting process. The neuro-optimization overcomes the shortcoming of low efficiency in conventional artificial intelligence. In the case of camshaft, the efficiency and precision of simulation are better in neuro-optimization than in FEM.