Virtual sample expansion method based on mechanical product historical data

Abstract

The invention discloses a virtual sample capacity expansion method based on the mechanical product historical data. The method comprises the following steps of firstly, determining the virtual samplecapacity of a virtual sample in a small sample problem according to the actually measured historical data; constructing a sample pool for generating the virtual samples based on the historical data ofthe mechanical production and the related priori knowledge of the mechanical production; then performing sample sampling based on a roulette sampling thought, and designing a virtual sample generation rule based on an agent model thought and a Jacobian spinor theory; and finally, reserving a feasible expansion sample through a sample rationality judgment condition, so that the small sample regression problem training virtual sample expansion for the mechanical assembly precision prediction is realized. The method can be used for expanding the sample capacity of a small-capacity sample machinelearning training model, can solve the problem that the number of the samples is insufficient in the mechanical assembly precision prediction, and has the important significance for researching the small samples for customizing the product tolerance transfer by using a machine learning regression method.

Description

technical field [0001] The invention relates to a method for expanding the capacity of a virtual sample, in particular to a method for expanding the capacity of a virtual sample based on historical data of mechanical products. Background technique [0002] Machine learning techniques based on large sample data have been widely used in different fields. With the introduction of the concept of intelligent manufacturing, the combination of machine learning technology and the mechanical field is becoming increasingly close. However, with the continuous improvement of mechanical design and manufacturing level, mechanical products have a trend of diversified design requirements, non-standardized and personalized customized production, which leads to some problems in the mechanical field no longer have the conditions to generate large-capacity sample data, thereby limiting application of machine learning techniques. [0003] At present, machine learning in the mechanical field is...

AI Technical Summary

Problems solved by technology

There are few studies on virtual sample generation methods for regression problems

Images