A shrinkage prediction method of investment casting based on geometric parameters

During investment casting process, the design of die cavity is the key factor to ensure the dimension accuracy of casting. The design principle of die cavity is based on computer-aided design model of casting. However, the casting will shrink in the die cavity with the decrease of temperature. Thus, the finalized design of die cavity is determined by the shrinkage and computer-aided design model of casting. It is necessary to predict the shrinkage of casting. A prediction method which based on adaptive learning idea of BP artificial neural network and construction of mapping model idea of regression method is proposed in this paper. The method is used for predicting shrinkage of casting. In this method, a series of I-beam castings that were divided into training samples and testing samples are designed for casting experiment and shrinkage measurement. Training samples are used for building shrinkage prediction model. Testing samples are used for verifying accuracy of the shrinkage prediction model. The results show that the shrinkage prediction method has higher fitting precision than regression method in fitting these training samples. Besides, for the training and testing samples, the prediction accuracy has positive correlation with fitting precision. The shrinkage can be well predicted by the prediction method through the known geometric parameters of computer-aided design model of casting.