Application Of Blind Source Separation In The Strain Rosette Offset Error Correction

As a kind of important industrial measuring element, the resistance strain gauge arewidely used in the field of productive and scientific research with the advantagesof lowerprice, reliable performance, and convenient operation etc. When the strain rosette is usedin measuring the principal stress of equipment, it is hard to make each strain gauge in anideal angle, and the deviation angle is usually unknown, which will affect the accuracy ofthe measurement results because of restriction from the complexity of equipment structure.Therefore, it is very essential to develop a method to reduce and eliminate the erroreffectively created by the patch offset of strain gauge.At present, the method to eliminate the patch offset error revises the deviation anglebased on the theoretical calculation, which is not only time-consuming and strenuous, alsorequires the technicist possessing higher professional skill. In this paper, a blind sourceslit-step algorithm in the correction process of the patch offset error of strain rosette isproposed, which can separate the measurement data and obtain the patch angle of eachstrain gauge. Thereby the principal stress is solved for achieving the purpose to reduce thepatch offset error.This paper analyzes the theory and method of solving rosette principal stress, a blindsource separation algorithm meets instantaneous linear mixed model is established, whichbased on normal strain and shear strain of each assume direction as the source signal andeach strain gauge measurement data as a mixed-signal.The proportional relationship between each element and the angle of the modelmixing matrix as a priori information is applied to the process of solving the blind sourceseparation, which solves the inherent uncertainties and identifies each strain gaugeplacement angle accurately.Experiments show that the angles identified by blind source separation algorithm ismuch closer to the real patch angle, the values of the principal stress and principal stressdirections are closer to the true value, which reduce the effects of the patch deviation erroron the measurement results and improve the test accuracy.