On the implementation of structure function in the digital signal processor

Abstract

The implementation of the structure function on a digital signal processor board to detect incipient stall and its applicability for online or near online applications were investigated. Previous research proved that the structure function successfully detects precursors to stall in axial compressors in off-line situations. The simplicity of the detection mechanism form a computational point of view suggests that the structure function could be used as an effective method for online stall detection. An eZdsp hardware platform from Spectrum Digital employing TMS320F2812 DSP by Texas Instruments was utilized to implement the structure function in this study. The ability to detect chaos in a time series was tested since the fundamental theory is affiliated with chaos theory. The experimental data obtained from high-speed compressors were then filtered with structure function implemented on the DSP board. The structure function applied to these systems showed sensitivity to the system dynamics and pre-filtering conditions. Due to this sensitivity the real time application is not feasible, although the near on-line detection is possible and has potential application in a supervisory control mechanism for stall warning.