Efficient measurement of aero-engine blade considering uncertainties in adaptive machining

Abstract

Blade geometry plays an extremely important role in the aerodynamic performance of aero-engine. However, in the adaptive machining process of a near-net-shape blade, it is important for an application engineer to reconstruct an actual profile with high-precision measurement data in a short time. This paper presents an efficient measurement to meet the requirement for digital information in the adaptive machining, which is also intimately associated with the previous work in reconstruction (Zhang et al. in Int J Adv Manuf Technol 2015:1–16, 2015). Comparing with conventional measurement, the proposed method can not only improve the utilization of measurement data from multi-sensor system but also consider the uncertainties of measurement and localization in data merging. Concretely speaking, the low-precision data from non-contact measurement device is used for establishing possible substitute models. The modified Bayesian information (Modi_BIC) criterion based on Bayesian statistics is introduced as the principle to select the model structure. In order to reduce misalignment errors of the selected substitute model, the statistical analysis of Fisher information matrix is applied to determine registration points. Based on this, the uncertainties can be given in the form of variation boundaries. Comparisons with the specified tolerance band determine a set of re-measured points by using a contact measurement device. The high-precision data can be merged with the selected substitute model by means of free form deformation (FFD) technique. The final measurement data is provided in the form of skinning surface that is consistent to the reverse modeling methodology of aero-engine blade. Moreover, some practical examples are given to demonstrate the effectiveness and superiority of the proposed method.