Comparison of DOA Algorithms Applied to Ultrasonic Arrays for PD Location in Oil

To reduce the number of elements and the cost of follow-up circuit of a $16 \times 16$ ultrasonic array sensor, two ultrasonic array sensors are designed. One is L shaped with 13 elements and the other is cross shaped with 25 elements. Based on these two ultrasonic array sensors, this paper compares the directional performance of a two-sided correlation transformation algorithm, which is a broadband focusing algorithm, an array expanded algorithm based on fourth-order cumulants, and a combination of these two algorithms, namely, synthesis algorithm. The broadband focusing algorithm can adjust to the broadband ultrasonic signal emitted by partial discharge (PD) source. After virtual expansion, the L-shaped and cross-shaped ultrasonic arrays can achieve the effects of an array with 97 and 193 elements, respectively. The adaptive capacity of broadband ultrasonic signal emitted by the PD source, and the precision of direction to PD are compared for the three algorithms. Directing simulation and experiments with single or multiple PD sources are carried out. The results show that the directing precision is highest with the synthesis algorithm, the error is <;2°, the peaks of spatial spectrum are sharp, and the best broadband adaptive capacity and noise suppression ability are achieved. This paper should provide a foundation for the application of array technology in enhancing the reliability of power transformer operation.