Aero-Engine Dynamic Start Model Based on Parsimonious Genetic Programming

Abstract

A novel parsimonious genetic programming (PGP) algorithm together with a novel aero-engine optimum data-driven dynamic start process model based on PGP was proposed. The method uses traditional GP to generate nonlinear input-output models that are represented in a binary tree structure. It introduces error reduction ratio (Err) to estimate the contribution of each branch of the tree, which refers to basic function term that cannot be decomposed any more according to special given rule. It applies orthogonal least squares algorithm (OLS) to eliminate complex redundant subtrees and then enhance convergence speed of GP. It is expected to obtain simple, reliable and exact linear-in-parameters nonlinear model via GP evolution algorithm. Application to real aero-engine start process test data validates that the proposed method can generate more robust and interpretable models. It is a rather promising method for complex nonlinear systems modeling with rather little prior system knowledge