Research Of On-line Control Method On Plastic Injection Molding Process Parameters

The craft parameters have direct influence on final part performance and quality in injection molding. The influence of disturbance makes it more difficult to get consistent good quality part. The investigation of on-line quality control for injection molding to make the molding keep on the desirable level and produce high quality part consistently is very valuable practically. In the formation process the online survey product quality condition is extremely difficult, because the judgment product quality (for example inspection product surface quality and survey size and so on) need the very long time, but online survey request judging product quality quickly in short time. In order to solve this problem, it is necessary to monitor the craft parameter indirectly. After the relationship between part quality indexes and contols parameters have been confirmed, the system can online control part quality by controlling craft parameters, and does not need examinating surface quality or survering part size.This dissertation take the process craft parameter as the object to carry on the dynamic modeling, on-line regulate the craft parameter based on dynamic rules and empirical rules, causes the product formation to be at the most superior condition, provide a closer application control model in order to reduce the ratio of disqualitied products and improve the part quality. The paper prime task includes:1,The experiment modeling based on Genetic Neural Network: Genetic Neural Network isadopted to establish the craft model between key parameters and qualitative indexes, because the injection molding formation is the complex multivariable, non-linearity also has the periodic non-stable state process. The craft model is known as product quality indexes predictor, it can forecast the products quality indexes according to the input parameters. In order to gain the training data based on Genetic Neural Network, the experiments are arranged on the plastic injection molding machine according to the improved orthogonal method, research the effects of key molding process variables(such as: melt temperature, injection pressure, injection speed and mold temperature) on the products quality. The author increase four group of supplement experiments according to the actual need. The compositor of various factors on influence degree is carried on through the analysis technology, and it is the foundation to establish the regulation order.2,Research of Genetic Neural Network: The coalescent of the genetic algorithm and the BP algorithm can be overcome the defect which the common neural network computational speed and overall optimization ability will give dual attention to with difficulty. The genetic algorithm can confirm quickly the region which the optimal solution is at, but it cannot gain the optimal solution soon. The BP algorithm has the relative stronger partial search ability but it' s convergence rate is slow. Therefore the coalescent of the genetic algorithm and the BP algorithm may mutually make up for one's deficiency by learning from others' strongpoint.3,Research of verification test: 16 group of verification tests has been arranged, the technical problem that how we can verify the accuracy of model and the adaptability of model can be resolved.4,Research of on-line control method on plastic injection molding process parameters: The craft parameter control model is established based on Genetic Neural Network which had been trained, this model can be used in the real-time product quality control in the actual production. Through studying expert knowledge and experience knowledge to confirm craft parameter adjustment direction and length of stride (that is experience rules). In the molding process, when the system detected critical injection parameters change and beyond the limited scope, it will launch the indicators to forecast the part quality. If the quality indexes overstep the scope of tolerance, the system initiate parameters adjustment procedure which based on dynamic rules and experience rules, identify new molding process parameters, so that our products always maintain the required quality limits.Through simply and directly model building and efficiently studying, the model which base on Genetic Neural Network can accurately reflect the varying relations between the key process variables and the product quality. At the same time, it was proved that the present method can adapt itself to all kinds of work conditions and has higher flexibility and aptitude. Moreover it can be used to predict the quality of products, process optimization and product quality control, so the model has high practical valueThe part quality control determines machine imports of the next cycle according product quality attributes of the previous cycle, adjusts machine setting and maintain the state to produce consistent, qualified products. The parameters control model which is presented in this paper can scout the state of cavity and achieve closed-loop control of the part quality .Implementation of this technology will greatly improve the product quality and quality consistency, and satisfy the higher requiremenand of part products, so it has very high practical value.