为了探究离心泵空化状态下声发射信号的频率分布特征及其变化规律，该文采用小波分析方法对离心泵空化试验过程中不同空化状态下的声发射信号进行了能量特征提取，研究了不同频段内声发射信号的能量特征随离心泵汽蚀余量变化的关系。试验分析结果表明，在离心泵的空化现象尚未出现以及不同的发展阶段，离心泵进、出口处声发射信号的全频段和各分频段绝对能量和相对能量的变化具有各不相同的特点。>31.25～62.5、>62.5～125和>125～250 kHz这3个分频段的绝对能量随空化状态改变而变化明显，且变化趋势与全频段的变化趋势类似。>31.25～62.5 kHz分频段的相对能量在空化状态变化过程中基本保持不变，而>62.5～125和>125～250 kHz 分频段的相对能量变化更为明显。研究结果可为离心泵空化状态的判别提供技术参考。%With the development of Acoustic Emission (AE) techniques, researches on the AE signals emitted during the cavitation process to judge whether the cavitation has occurred or not and its development degrees are getting more and more attentions. However, the frequency distribution features and their variation laws with the cavitation states of centrifugal pumps remain to be further studied. In this paper, the wavelet energy features of the AE signals under the cavitation conditions of a centrifugal pump were extracted with the wavelet analysis method, and the variation laws of these features were also revealed. Firstly, the cavitation processes of the centrifugal pump with constant flows and constant speeds at the 60%, 100%and 120%rated flow respectively were simulated on an open centrifugal pump test bench. Each cavitation process included such different stages as no cavitation, cavitation just appearing and cavitation developed with different degrees. Simultaneously, the corresponding operational parameters of the centrifugal pump and the AE signals emitted near the inlet and outlet casing of the centrifugal pump were collected. Based on the calculation of the heads H and the net positive suction heads NPSH, the cavitation curves with different flows were obtained. Secondly, the AE signal energy feature extraction method based on wavelet analysis was studied, so as to obtain the specific frequency bands’ absolute energies and relative ones. Then, taking the 60%rated flow as an example, five typical operating points on the curves were selected and marked as A, B, C, D and E, representing different cavitation stages. The five-layer decomposition and the soft threshold denoising with the mother wavelet db5 were implemented on the AE signals near the inlet and outlet casing of the centrifugal pump corresponding each operating point. These AE signals were divided into six frequency bands, i.e. 0-31.25, >31.25-62.5,>62.5-125, >125-250, >250-500 and >500-1 000 kHz, and the absolute energy and the relative one in each band were calculated. Finally, the relation curves of these energies with NPSH from the operating point A to E were compared and the variation rules were analyzed in details. The results showed that the AE signals’ absolute energies in the whole frequency band near the inlet and the outlet casing of the centrifugal pump varied both obviously with the changes of NPSH. With the cavitation developing from nothing to serious degrees, the NPSH values remained unchanged firstly then fell sharply, after that increased gradually and finally closed to the values before the occurrence of cavitations. However, the changes of the AE signals’ absolute energies in the six frequency bands were not identical. The absolute energies in 31.25-62.5,>62.5-125 and>125-250 kHz varied obviously with the cavitation stages in the same way to the absolute energy in the whole frequency band, while those in the 0-31.25,>250-500 and>500-1 000 kHz bands changed very little. It can be inferred that the 0-31.25 kHz band corresponded to the background noises due to the fluid motion and the interaction between the fluid and the mechanical components of the centrifugal pump, and the latter two higher frequency bands corresponded to the electromagnetic noises and other random noises in the test field. The components of these three frequency bands were irrelevant to the cavitation states. Therefore, the absolute energy changed in the frequency bands such as 31.25-62.5,>62.5-125 and>125-250 kHz can reflect the cavitation states directly, which should be paid more attentions. Moreover, the relative energies in the 31.25-62.5 kHz band for both AE signals kept unchanged during the cavitation process, while those in the 62.5-125 and>125-250 kHz bands varied obviously. These wavelet energy variation laws can serve as a technical reference for identifying cavitation stages of centrifugal pumps.
展开▼
