Foundry technology parameter optimization method based on casting defect prediction model

Abstract

The invention belongs to the related technical field of foundry technology optimization and discloses a foundry technology parameter optimization method based on a casting defect prediction model. Themethod comprises the steps that defect information reflecting the technology level of a casting as a detected object is automatically extracted, and then defects are subjected to qualitative classification and quantitative management; a data package composed of three parts of multi-source data including production data, technology data and detection data is collected and stored; based on the datapackage, a depth neural network is adopted for establishing and optimizing the casting defect prediction model; an orthogonal test is designed to input a technology parameter combination into the model, a sensibility mechanism of key parameters for hole loose defect evolution is researched, and a key technology parameter optimization window is acquired. By means of the method, the intelligent prediction process of the casting defects can be executed, the multi-source data in the whole foundry technology can be related and applied, the optimization result is obtained by a higher-precision modecorrespondingly, and accordingly the quality of the final casting is improved.

Description

technical field [0001] The invention belongs to the related technical field of casting process optimization, and more specifically relates to a casting process parameter optimization method based on a casting defect prediction model. Background technique [0002] As a widely used metal thermal processing process, casting has a wide range of raw materials and strong adaptability. The product range involves many fields, such as electronics, chemicals, medical equipment, weapons, aviation, etc., which have an important impact on human production and life. The quality of castings is closely related to the level of casting technology, and the level of technology directly determines the quality and safety of castings. In general, the casting process can be divided into three basic stages, namely, the casting metal preparation stage, the casting mold preparation stage and the casting processing stage. Due to the many processes involved in the whole process and the complex parameter...

AI Technical Summary

Problems solved by technology

For example, CN201710684770.6 discloses a method of applying BP neural network to predict shrinkage cavity defects of TC4 titanium alloy castings, in which the difficulty in detecting shrinkage cavity of titanium alloy castings is solved by applying the BP neural network model to the prediction process of casting shrinkage cavity position , and the accuracy of casting model software prediction results is insufficient

However, further studies have shown that the above-mentioned existing scheme still has the following defects or deficiencies: First, this scheme fails to take into account more influencing factors in the entire casting process, and often only determines the shrinkage type of defects , it cannot comprehensively and accurately reflect the comprehensive quality problems of castings in actual work; secondly, the BP neural network algorithm it chooses cannot deal with nonlinear problems, and its ability to classify, identify and process multi-source data is not enough, which also affects the prediction results efficiency and accuracy; finally, the effective use of forecasting results is still insufficient

Images