摘要：AbstractThis study presents a method to reduce the number of scheduling variables in a linear parameter-varying (LPV) model of a diesel engine air path system. The reduction of these scheduling variables is very important because it exponentially decreases the computational complexity for the gain-scheduled LPV controller synthesis. Principal component analysis (PCA) and autoencoder (AE) based neural networks are applied to the LPV diesel engine’s air path model, and the relationship between the accuracy of the reproduced scheduling variables and the number of the reduced scheduling parameters is evaluated via conduction of numerical simulations.
关键词：KeywordsDiesel engineair path systemLPV modelautoencoder neural networksprincipal component analysisgain-scheduled control
