Pseudo-multi-source-sampling complex domain FastICA for impact location

Using the sensor array to locate impact can help to detect damage and its potential threats for structure materials, and to ensure structural safety. Sensor array signals are preprocessed by blind source separation (BSS). A single-channel pseudo-multi-source-sampling method is proposed, and data sequences of one sensor in different periods are used as multi-source observation signals of this sensor. The signals of all sensor arrays are used as inputs of complex domain FastICA impact extraction algorithm. Shock signals with phase information are separated from each sensor. The coordinates of impact events on test plane are solved by using the threshold time delay localization method. theoretical analysis and numerical simulation verify the effectiveness of the combined impact location algorithm. Experiment on the impact location platform shows that designed algorithm can separate out the impact from the signals mixed with vibration. It is suggested that the algorithm improves the precision of time delay analysis for impact location in the sensor array. It also provides a practicable solution for the underdetermined blind source separation problem while the number of sensors is less than the sources.