Blind Signal Separation Method Based Machining Error Decomposition

In view of the current error separation method cannot successfully separate the dissimilar systematic error components from the whole surface machining data, a method for machining errors decomposition based on blind signal separation is proposed which can distinguish those dissimilar systematic error components. Firstly, the error transfer model to describe the errors synthesizing from the single error component caused by corresponding individual error source to the synthesis machining surface error is conducted. To determine the number of the systematic error components, the principal component analysis (PCA) method is used. Then according to the theory of blind source separation, negative entropy based fixed point algorithm is proposed to fulfill the machining error components separation, which can realize the separation of the systematic error components even in close frequency scale. Finally, a shaft surface finish turning data and a flat surface milling data are used as examples to verify the proposed method. The result shows that the proposed error separation method can effectively realize the machining error separation in close frequency scale.