Study On The Characterization Method Of Cavitation State In Pumps Based On Sound And Vibration

As a common phenomenon in pumps,cavitation leads to a series of detrimental consequences,including degradation in hydraulic performance,erosion of over-current components,violent noise and vibration,etc.It is a crucial limiting factor to the efficiency,safety,and stealthiness of pumps.The prevention and control of cavitation damage are significant,and cavitation characterization based on cavitation feature extraction has a high potential to deal with this demand.In this paper,the cavitation experiments of pumps are conducted,the signal model of cavitating pump is established,and the methods of signal demodulation,qualitative identification,and quantitative characterization of cavitation in pumps are proposed.The main research contents are summarized as follows:1.The cavitation experiments of centrifugal pump and waterjet pump under visible conditions are implemented,and the fluid field images and acoustic and vibration signals are acquired synchronously.On this basis,the evolution process of cavitation is summarized.The results show that cavitation in centrifugal pump mainly exists in the form of attached cavity,while cavitation in waterjet pump includes unattached cavity and attached cavity.Unattached cavity is of higher stability than attached cavity,thus presenting lower characteristic frequency.In contrast,attached cavities such as sheet cavity and cloud cavity are relatively unstable,which leads to higher characteristic frequency.As the cavitation state of pumps develops,the physical behavior severity of cavities gets higher(e.g.,oscillation,pulsation,shedding,and collapse),and the spatial distribution of cavities becomes more unstable.2.The cavitation modulation mechanism leaded by the synergy between cavitation behavior and impeller rotation is investigated.The excitation sources and statistical features of the acoustic and vibration signal components of pumps are sorted out.Afterwards,the signal model of cavitating pump is established in the framework of cyclostationary processes.The results show that the acoustic and vibration signal components consist of modulation components caused by cavitation,cyclostationary noise,and stationary noise.Both amplitude modulation and frequency modulation are involved in the cavitation modulation mechanism.Cyclostationary noise mainly results from mechanical excitations and fluid phenomena except cavitation,while stationary noise comes from the environment and the acquirement system.The information of modulation components caused by cavitation includes two aspects: the modulation function reflecting the periodic rotation of impeller and the carrier frequency distribution revealing the physical behaviors of cavities.3.The research on extracting characteristic frequencies by demodulation techniques is carried out based on the cavitation modulation mechanism.The time-domain statistical analysis of second-order cyclostationary components is implemented,and the Enkurgram is proposed in the framework of narrowband demodulation.The frequency-domain statistical analysis of second-order cyclostationary components is conducted,and the synchronous modulation spectrum is proposed based on cyclostationary theory.The proposed methods are verified by simulation signals and actual data and compared with the classical narrowband demodulation methods and spectral quantities of cyclostationary theory.The results show that the Enkurgram is capable of extracting modulation frequencies effectively in the situation of heavy Gaussian and impulsive noise.In contrast,the synchronous modulation spectrum is able to focus on extracting the target modulation frequencies under the interference of non-cyclostationary and cyclostationary noise.4.The diagnosis strategy of qualitative cavitation identification based on carrier distribution is constructed.Under this guideline,the method of Carrier distribution Estimation based on Time-Frequency Analysis(CE-TFA)and the method of Carrier distribtuion Estimation based on Cyclo Stationary Analysis(CE-CSA)are proposed.These methods are verified by simulation signals and actual data.Afterwards,the procedure of intelligent recognition of cavitation state in pumps is designed by combining time-frequency analysis,demodulation analysis,and Deep Convolutional Neural Network.The results show that the CE-TFA is able to realize the enhanced estimation of cavitation carrier distribution,while the CE-CSA is capable of estimating cavitation carrier distribution with good anti-noise performance.Compared with the classical time-frequency analysis such as Short-Time Fourier Transform and Wavelet Transform,the CE-TFA and CE-CSA show better performance.Through the proposed methods,the qualitative identification of cavitation type can be realized,and even the characterization of cavitation degree can also be achieved when the energy of cavitation modulation components changes significantly during the evolution process of cavitation.5.The diagnosis strategy of quantitative cavitation characterization based on modulation intensity indicators is constructed.Three modulation intensity indicators—Absolute carrier Power(AP),Relative carrier Power(RP),and Characteristic Modulation Components Ratio(CMCR)are designed to evaluate the overall behavior strength of cavities,the Signal-toNoise Ratio(SNR)of cavitation modulation components,and the spatial distribution stability of cavities,respectively.The proposed indicators are verified by simulation signals and actual data.The results show that the AP proves to be useful in measuring the early developing cavitation,the CMCR is capable of detecting the key turning point from the early developing cavitation to the fully developed cavitation,and the RP can reflect cavitation degree from the view of SNR.In summary,these indicators solidly complement each other.Thus,their combination provides an efficient solution to cavitation quantitative characterization.This paper summarizes the evolution processes of cavitation in centrifugal pumps and waterjet pumps and establishes the cyclostationary signal model of cavitating pumps.On this basis,the diagnosis strategies of signal demodulation based on cavitation modulation mechanism,qualitative cavitation identification based on carrier distribution,and quantitative cavitation characterization based on modulation intensity are designed.Correspondingly,the demodulation methods,including the Enkurgram and the synchronous modulation spectrum,the carrier estimation methods,including the CE-TFA and the CE-CSA,and several modulation intensity indicators are proposed.The above research works can provide the theoretical fundamentals for the stable,reliable,and high-efficient operation of pumps in industrial applications.