喷水推进泵空化特征联合分类识别

在无空化区、空化Ⅰ区和空化Ⅱ区3种工况条件下，利用安装在实船喷水推进泵检查孔内的水听器和布置在泵壳上的加速度传感器获取喷水推进泵的声压和振动信号.通过二阶导数分析发现，空化前后声压和振动信号的二阶导数均具有显著的差别，空化后声压信号在8~25 kHz、振动信号在3~12 kHz频段上出现了明显的峰值.以该特征频段上二阶导数的峰值序列为基础，构造特征集并进行特征联合分类识别.结果表明：利用特征联合分类识别可以区分无空化区、空化Ⅰ区和空化Ⅱ区的信号，多传感器特征联合分类识别的效果优于单传感器特征联合分类，尤其是基于二阶导数的多传感器特征联合分类识别的效果较为理想.

Features Combined Classification of Cavitation in Waterjet Pumps

Onboard experiments were carried out when the waterjet pumps are working on the conditions of none cavitation zone, cavitation zone Ⅰand cavitation zone Ⅱ, acoustic and vibration signals were acquired respectively by hydrophones mounted in the inspection holes and acceleration transducers mounted on the case of the waterjet pumps. It showed that both of the second derivatives of the acoustic and vibration signals have notable differences before and after cavitation, and visible peak values appear in 8-25 kHz of acoustic signal and 3-12 kHz of vibration signal after cavitation. Cavitation feature sets were constituted based on the peak series of the second derivatives for features combined classification. The results show that the three types signals can be distinguished well by features combined classification, and the effect of multi-sensors features combined classification is better than that of single-sensor, especially the effect of multi-sensors features combined classification based on high order derivative is more perfect.