3D-intensity measurements may be used as a straightforward method for source localisation. In reality, however, there is a complex relation between the near field intensity estimate and the sound pressure level in the far field. Anothercomplication is that the estimated intensity may be a combination of direct radiation and interference from coherent secondary sources [1].The aim of this study is to enhance interpretation of FFT-based intensity measurements by using principal component analysis (PCA) [2-4], for post processing of the data. PCA of intensity data resolves correlated patterns in both frequency and spatialdomain. The advantages are that non correlated noise is removed and that original data is decomposed in principal components which may be related to independent sources [5]. To verity the method, a model experiment was performed. The objectives are: (1)to increase the accuracy of the vector estimate, (2) to decompose the sound field in independent components and thereby identity inherent noise sources, (3) to see in what extent partly coherent sources can be separated.
展开▼
