Quality Control Of Small-Batch Production Based On The Improved Grey RBF Neural Network Model

With the increasingly competitive global market, product quality gets unprecedented attention in enterprises. In addition, with the diversification and personalization of customer’s needs, the mass production of traditional model has been replaced by multi-varieties&small batch production. However, small batch production of the product less, fewer samples, therefore the traditional method of quality control is difficult to be applied directly in the small-batch production model. In particular, for the higher accuracy requirements of the productions, we need more effective control method. Therefore, how to apply improved existing method in controlling process of multi-varieties&small batch and put forward a more effective method becomes an urgent task.Firstly, the paper briefly describes characteristics of multi-varieties&small batch production. On the basis, the traditional quality methods and modern intelligence methods of quality forecasting are introduced and classified in this paper. At the same time, several major methods have been selected to do a detailed analysis on the scope of the advantages and disadvantages of each method.Secondly, through the gray system theory and RBF neural network, the establishment and application of the two algorithms are mainly introduced in the quality predictive control. An improved model is established based on the traditional GM(1,1), which can select optimal background value and dynamic identifying parameters and is fit for the prediction of small sample data.Finally, in this paper, we combine improved GM (1,1) and RBF neural network and use the advantages of both to build grey compensating RBF neural network forecasting mode. In order to verify its effectiveness, the model was applied into the selected example. Then compare the predicted results to dynamic exponential smoothing and unimproved RBF neural network predicted results. The calculation proves that optimized prediction model’s accuracy and stability are higher than the two contrasting model.Through the research above, a grey compensating RBF neural network forecasting model was proposed by combining grey theory and RBF neural network. It can control the quality characteristics of multi-varieties&small batch product in advance and improve quality prediction accuracy, reducing the probability of non-conforming product. So enterprises can enhance its competitiveness and achieve better efficiency by using this method. In a word, the research of improved gray neural network model has certain theoretical and practical significance to quality control in multi-varieties&small batch production.