The Research Of Injection Molding Processing Optimization Based On Hybrid Intelligent Technology

Injection molding is an efficient molding technology for plastic products. With the popularization of plastic products used in industry, its virtue and importance is standing out increasingly. But non-linear and uncertain relation between processing facters and parts quality makes the processing optimization and quality control very diffcult. Traditional method of regulating the processing is not only time-consuming and high-cost, but also excessively dependent on experiences or sucessful cases. And its accuracy is not high. It is not suitable for the short term reqirement for new products producing especially. Although numerical simulation can reduce the cost and give the qualitative guidance for processing regulating, it lacks quantitative precision. Furthermore, the method is confined to the description between one factor and one target. Experiences and experiments are still essential to obtain a good testing result. Good design of experiment can reduce the blindness of trial and error to some extent and obtain the good combination of processing variables within the testing scope. But limited values of per variable studied in the experiments is usually not able to be the optimal solution in full.Many scholars have recently tried all kinds of intelligent technology boldly and acquire some research fruits. The majority of them either rely on expert knowledge and intervening, or only study one index such as surface roughness, key dimensions and warpage, or make qualitative researchs. Until now there have been few researches about comprehensive optimization of injection molding quality. So it has particularly far-reaching meaning to study in this field while current market concerns quality and efficiency more greatly than before.Several methods of processing control are narrated in detail, combined with the nowaday research status of injection molding processing after rheology behavior of polymeT and the numerical simulation theory of thermoplastic are discussed. Using the successful cases for reference, the optimization idea of hybrid intelligence is put forward in which intelligent technology is introduced into the optimization, neural network, genetic algorithm and fuzzy comprehensive valuation are combined together on a base of the sample data from numerical simulation orthogonal experiment. The study of a deep-cavity and shell-shape part has proved the idea feasible finally.The main content of the thesis is as follows:1, The Theological theory of injection molding processing and effect of processing parameters on parts qulity are discussed deeply at first. And on the base of the current CAE software of injection molding the relative simulation theory is studied. Than in consideration of the design and use requirments of parts, quality index and relevant processing variables are determined.2> According to the theory of design of experiment, all the experiments is scheduled and experimental data are obtained by numerical simulation software. Finally, the relation between variable and index is studied within the range of research on the analytic result of data statistic.3^ BP neural network is created to predict the parts quality. The non-linear mapping relation between variables and index is obtained by using data of orthogonal numerical simulation experiment as sample data, designing the structure of the network, selecting the learning algorithm and iteration optimization. This stucture of BP network serves the computation tool for the optimization below. The intelligent optimization of parts quality is realized by citing the thought of natural evolution into the genetic algorithm, using fuzzy comprehensive valuation formula as the fitness function and processing parameters as the design variable, taking advange of approximate computation of BP network above. In contrast with the results of numerical simulation orthogonal experiment, it is proved that the intelligent optimal method can find the optimal solution rapidly, exactly and evidently.The idea of hybrid intelligent optimization is not only fit for the processing optimization of injection molding, but also for optimization of other problems with multi-facter effect, especially multi-index restriction, non-linear and uncertain relationship. The whole course of searching solution needs no intervention, no apriori knowledge, so it has a great part in trial-producing of new products.