Error prediction for machining thin-walled blade with Kriging model

Abstract

Compressor blades are the key components of aero-engines, and their machining accuracy is critical to aero-engine performance. However, the choice of machining parameters during machining compressor blade has a direct impact on the position error and torsion error, which in turn affects the aero-engine performance. The conventional methodology for investigating mechanisms on machining error is frequently both time-consuming and labour-intensive. The agent-based modelling approach is trained on a limited set of experimental data in order to obtain an approximate mathematical model of the real process. This approach has the advantages of low modelling cost, convenient operation, and high computational efficiency. Accordingly, this paper employs the agent model to construct prediction models for the position error and torsion error of compressor blades. Firstly, experiments were designed to be conducted on compressor blades under different working conditions in order to obtain the position error and torsion error data of compressor blades. Then, based on the superiority of the agent model, the Kriging models are constructed to establish prediction models for the position error and torsion error of compressor blades. Finally, the influence of machining parameters on the position error and torsion error of compressor blades is analysed. The prediction accuracies of the established models are all greater than 0.85, which can provide strong support for the optimization of the machining process of compressor blades.