| RTM is a kind of multi-phase, multi-scale, multi-variable, non-steady-state, non-isother-mal and non-linear complex molding process with a certain periodicity. The influencing factors on the quality of the RTM molding are very complex, and can generally be sorted into four types:material parameters, machine parameters, process parameters and the disturbances. Molding process parameters are the most important influence factors, which directly determine the forming quality of a RTM process, the mechanical property and the usability of the final products.Choosing process parameters by experience has such stong subjectivity that the results are always unreliable. It’s possbile to obtain the exact solution of the simple model using the theo-retical method, however hard to dispose the complicated engineering practical problems, for example, the mold-filling process of non-newtonian fluid under the non-isothermal conditions, because the governing equation of which is a higher order nonlinear spartial differential equa-tion. Simarly, numerical calculation is more suitable for dealing with simple engineering prob-lem, and when the models or boundary conditions are sophisticated, it will be difficult to ex-press the physiochemical process using the numerical language and to write the program. The experimental method is cost-ineffective because a large amount of resources, energy and man-power will be consumed and the optimization cycles are usually long. Some optimal design methods, such as the genetic algorithm (GA), the particle swarm optimization (PSO) and the ant colony algorithm (ACO), frequently encounter premature-convergence, misconvergence or slow-convergence phenomena. On account of those problems, an improved genetic algorithm for optimizing injection pressure was proposed maily on the basis of rough set theory in the third chapter of this article. The discernibility matrix method was used to reduct the attributes of a RTM injection pressure knowledge-base. Then, a similarity calculation method was de-signed to extract the specimens from the knowledge-base as initial population. After the injec-tion time was choosed as the objective function, and the fitness function was projected corre-spondingly, a series of general genetic operations were conducted in the following.The case indicates that more than90%specimens extracted from the knowledge-base after reducting are the same with that before reducting, however the retrieval speed after reduction rose57%than before, which illustrates that attribute reduction not only do no harm to the knowledge classifi-cation, but improve the computing speed significantly. Moreover, the evolution curve shows that improved GA converged at the eighteenth generation, while the standard GA converged at the twenty-fifth generation, which demonstrates that the convergence speed of the improved GA is faster.Traditional theoretical methods are over-dependent on empirical constants with no physi-cal meanings. Almost all of the existing researches on permeability are about plane and twill fabrics, and few about satin fabrics. As to those problems, a new fractal permeability model was projected on the basis of the fractal theory, minimum potential energy principle, and Darcy’s law. Repetitive unit should be selected first according to the structural characteristics of fabrics. The tortuosity fractal dimension of pores can be figured out after the actual length of a pore calculated in accordance with minimum potential energy principle. The equivalent di-ameter of the maximum pore, on which one can work out the structure fractal dimension, can be obtained by microscopic measurement. Finally, the permeability can be expressed as the explicit equation of various structural parameters, such as, the area of cross-section, representa-tive length, equivalent diameter of the maximum pore, tortuosity fractal dimension and struc-ture fractal dimension, which makes the physical meanings intuitive and clear. The permeabil-ities of a5-harness and an8-harness were predicted according to the new fractal model, and a group of contrast experiments were carried out. Results showed that the predicted values agree well with the measurements, which illustrates that the new model is feasible and effective.Researches found that the optimization results are trustless when choosing shorest flow path as the goal to optimize the gate location, and the optimization accuracy of the existing simulation softwares needs to be improves. As to those problems, a new gate location scheme evaluation criterion (or mold-filling quality evaluation function) was put forward in the fifth chapter. Based on the evaluation function and the finite element/control volume method, a gate location scheme evaluation system was developmented, and eventully realized using the object oriented language-Delphi. In the end, a wing of an unmanned aerial vehicle (UAV) was analyze using the new method and a group of contrast analysis was setted. The preferred scheme filtered by the new criterion is scheme7, however the the preferred scheme filtered by the criterion of shortest flow path is scheme3. Simulation results indicate that though the filling time of scheme7is longer than scheme3, the varance of the pressure, shear stress, warpage, volume shrinkage and sink index of scheme7are smaller than than of scheme3, which means the filling quality of scheme7is great than that of scheme3, and the proposed gate location scheme evaluation function is credible. |
