Quality prediction method based on error propagation network and boosting tree algorithm

Abstract

The invention discloses a quality prediction method based on an error propagation network and a boosting tree algorithm. The method comprises the following steps: building a manufacturing resource relation network based on part machining features and processing elements; constructing a multi-procedure error propagation network combining the manufacturing resource relation network and a quality sub-network; determining input and output features of a quality prediction model, and constructing a quality prediction model based on an error propagation network and a boosting tree algorithm; respectively optimizing hyper-parameters by utilizing a particle swarm optimization algorithm and a grid search algorithm; establishing evaluation indexes of model accuracy and maturity; and speculating the product percent of pass by using the production field simulation data generated by using a monte carlo method. According to the method in the invention, visual modeling of the product production process is realized, and a quality prediction method which is stable in prediction capability, convenient to optimize in parameters and high in efficiency and accuracy is designed, so that the product quality can be accurately predicted by enterprises, the processing quality can be prevented and controlled in advance, the enterprise quality loss can be reduced, and the economic benefits can be improved.

Description

technical field [0001] The invention belongs to the field of processing quality prediction, and in particular relates to a quality prediction method based on an error transfer network and a lifting tree algorithm. Background technique [0002] The manufacturing quality of products is comprehensively affected by various factors such as people, machines, materials, methods, environment, and testing, and the influencing process is a complex nonlinear process. At present, enterprises mostly use SPC and other process state monitoring methods to control the processing quality of parts. Through the abnormal judgment of the model diagram, the abnormal production conditions in the processing process are fed back, and then the abnormal conditions are dealt with. This method of parts processing quality control It belongs to post-event control, which brings the risk of unqualified product quality to the enterprise. At present, most of the quality prediction methods based on intelligent...

AI Technical Summary

Problems solved by technology

The range of hyperparameters that need to be considered in SVR prediction is large, the optimization process is complex and easy to fall into local optimum

At the same time, under complex working conditions with multiple processes, both of them have the problem of limited prediction ability and low prediction accuracy.

Images