A Method of Building a Surrogate Model for Complex Product Optimal Design Based on Small Samples

Abstract

A method for establishing an agent model for complex product optimization design based on small samples, belonging to the technical field of complex product optimization design. Determine the goal of the complex product optimization design; generate an original design solution sample set with a sample size S for the complex product optimization design; generate a virtual design solution sample set for the complex product optimization design; merge the original design solution sample set and the virtual design solution The sample set constitutes a sample set of mixed design solutions; determine the sensitivity and sensitivity ranking of the goals of complex product optimization design relative to each decision variable; use the target variable as the output variable to establish a three-layer BP neural network model with different input variables; use mixed samples The set of training samples is used to train each of the above neural network models; the neural network model with the best performance is selected as the final agent model for complex product optimization design. It reduces the workload of generating samples and ensures the accuracy of complex product optimization design proxy models.

Description

technical field [0001] The invention belongs to the technical field of complex product optimization design, in particular to a method for establishing a complex product optimization design agency model based on small samples. Background technique [0002] Complex products refer to a class of products with complex customer needs, complex product composition, complex product technology, complex manufacturing process, and complex production management. Aircraft, engines, ships, and machine tools are typical representatives of such products. Complex product optimization design is an optimization process that continuously adjusts design parameters to form a new solution, and evaluates whether the adjustment is effectively approaching the design goal. This is a process of trial and error. If the original design method is used for the effectiveness evaluation of each tentative adjustment, it will bring unacceptable amount of calculation, which will eventually lead to the infeasibil...

AI Technical Summary

Problems solved by technology

However, the technical complexity shown in the complex product design process leads to the high cost of generating design samples, and it is difficult to obtain enough training samples, which leads to the inability to estimate some feature extraction parameters and the inability to establish an accurate proxy model; in addition, how to fully Considering the high-dimensional problems brought about by technical complexity, choosing a cost-effective model will fundamentally determine the overall quality of the optimized design

Images