The machining error control of blade shape based on multivariate statistical process control

Abstract

Due to the high demand on blade shape quality and accuracy, it is essential to control the quality fluctuation in blade machining process. Since there are multiple correlative quality characteristics on blade shapes, multivariate statistical process control is adopted to monitor blade quality fluctuation. Meanwhile, blades, which are designed to satisfy better aerodynamics, always have complex and special shape, and the high dimensionality problem hinders the implementation of multivariate control charts. In order to deal with this problem, blade shape profiles are first divided into several segments according to the geometrical shape and the variable tolerance requirements of blade section curves, which are monitored simultaneously. Then, the quality control vector can be built, which comprises profile error and bending–torsion deformation. Because of the high dimensionality of data points on blade section curves, the equal error discretization method is used to deal with the error samples in a computationally lightweight manner. Due to the requirement for high machining accuracy, these proposed statistics that form the quality control vector can be monitored by a small shift multivariate control chart such as the multivariate exponentially weighted moving average control chart. The mean distance difference vectors for each segment, which are designed into the multivariate exponentially weighted moving average control chart, can not only detect shifts but also provide diagnostic information when a blade shape profile is out of control. Finally, the validity of the proposed method is proved by a simulation study applied to a blade machining process.