A pure water hydraulic pump valve vibration monitoring method and system

Abstract

The application provides a pure water hydraulic pump valve vibration monitoring method and system, and relates to the technical field of pump valve vibration monitoring.The method comprises the following steps: obtaining a high-frequency characteristic signal and a low-frequency characteristic signal; identifying a high-frequency transient event according to the high-frequency characteristic signal, and recording the time point and transient characteristic of the high-frequency transient event; extracting a low-frequency signal segment synchronized with the high-frequency transient event from the low-frequency characteristic signal according to the time point of the high-frequency transient event, and determining the impact vibration characteristic of the low-frequency signal segment; and determining the cavitation state of the pure water hydraulic pump valve according to the transient characteristic of the high-frequency transient event and the impact vibration characteristic of the low-frequency signal segment.The method aims to solve the technical problem that cavitation signals are complex and superimposed in the operating environment of the pure water hydraulic pump valve, and are difficult to separate and identify, and to realize accurate and robust identification of early weak cavitation phenomena, so as to avoid false negatives and false positives, thereby effectively warning and evaluating cavitation damage.

Description

Technical Field

[0001] This invention relates to the field of pump and valve vibration monitoring technology, and more specifically, to a method and system for monitoring the vibration of a pure water hydraulic pump and valve. Background Technology

[0002] In industrial production, pure water hydraulic systems, with their clean and pollution-free characteristics, are used in fields such as semiconductors, food and pharmaceuticals, and precision machining, where the purity of the medium is strictly required. This system uses pure water as the power transmission medium to drive pumps, valves, and other actuators to complete operations. Pure water differs from traditional hydraulic oil in its physical properties, such as lower viscosity, lower density, and higher saturated vapor pressure. These characteristics make pure water prone to vaporization when flowing through areas where local pressure is momentarily reduced, such as pump inlets or valve throttling orifices, forming bubbles. When these bubbles move with the fluid to high-pressure areas, they rapidly collapse, generating high-intensity shock waves and microjets—a phenomenon known as cavitation. The long-term action of cavitation shock waves on the metal inner walls of pumps and valves can lead to pitting and peeling of the material surface, forming cavitation erosion damage. Accumulated damage can cause a decline in component performance, reduced efficiency, and even failure, affecting production line operation.

[0003] To provide early warning of cavitation damage, acoustic emission (AE) detection technology has been introduced. Acoustic emission refers to the phenomenon where a localized, rapid release of energy occurs within a material, propagating outward as high-frequency elastic waves. The collapse process of cavitation bubbles is a typical source of acoustic emission, generating signals with frequencies typically ranging from tens of kilohertz to several megahertz, higher than the frequencies of conventional mechanical vibrations. Therefore, by placing high-frequency acoustic emission sensors outside the pump / valve housing, the high-frequency signals generated by cavitation can be captured.

[0004] However, in practical pure water hydraulic systems, the detection of acoustic emission signals faces multiple complexities. When cavitation bubbles collapse in pure water, they release instantaneous high energy, which spreads outwards as high-frequency pressure waves. The frequency range of these pressure waves is typically from tens of kilohertz to several megahertz, and their propagation characteristics are affected by the physical properties of the pure water medium itself. Compared with traditional hydraulic oil, pure water has low viscosity and density, as well as high sound velocity, which makes the propagation attenuation characteristics of cavitation-generated sound waves in pure water different from those in other media. More importantly, the internal structure of pure water hydraulic pumps and valves is complex, including components such as the pump impeller, volute, plunger cavity, valve seat, valve core, and throttling orifice. These components form tortuous flow channels and cavities. When cavitation sound waves propagate through these complex internal structures, they frequently encounter interfaces (such as water and metal walls, water and moving parts surfaces), resulting in multiple reflections, refractions, scattering, and diffractions of the sound waves. This multipath propagation effect not only causes sound wave energy attenuation, but also distorts the waveform and frequency components of the original cavitation signal, producing interference and standing wave phenomena. This results in uneven distribution of effective sound energy reaching the inner wall of the pump valve, and its characteristics are no longer those of the pure original cavitation signal.

[0005] These distorted sound waves propagating in pure water must then cross the water-metal interface to reach the pump / valve housing. An acoustic impedance mismatch exists between the water and metal, causing most of the sound energy to be reflected at the interface, with only a portion penetrating into the metal housing to continue propagating. The frequency components of the sound waves penetrating into the metal housing may couple with the inherent vibration modes and resonant frequencies of the pump / valve housing itself. This coupling causes the frequency components of the original cavitation acoustic emission signal to be modulated by the vibration of the housing structure, or to generate harmonics and subharmonics. Therefore, the signal received by the sensor is not a pure cavitation acoustic emission signal, but rather a superposition of the cavitation signal and the structural vibration response.

[0006] Furthermore, during operation, the internal mechanical moving parts of the pure water hydraulic pump valve (such as the rotation of the drive motor, the periodic meshing motion of gears or plungers inside the pump, and the reciprocating motion of the valve core) also generate mechanical vibrations. These mechanical vibrations are transmitted to the housing through the structural components of the pump valve (such as bearing seats, flanges, and connecting bolts). The frequency components of these non-cavitation-induced structural vibration signals may overlap with the low-frequency part of the cavitation acoustic emission signal, or their high-frequency harmonics may be close to the frequency of the cavitation signal, increasing the difficulty of signal identification. Acoustic emission sensors are usually installed on the outer surface of the pump valve housing. The installation position of the sensor and the coupling method with the housing (such as using coupling agent or bolt fixing) directly affect the signal pickup efficiency and fidelity. Poor coupling can lead to signal attenuation or distortion. At the same time, the pump valve system is connected to the entire mechanical system through pipes, supports, etc. These connection points become propagation paths for mechanical vibrations and noise, transmitting structural vibrations from inside the pump valve (caused by cavitation excitation or non-cavitation factors) and mechanical noise from the external environment (such as motor vibration and vibration of other equipment) to the surface of the sensor housing.

[0007] In summary, the signal ultimately received by the sensor is a mixture of multiple signals: cavitation acoustic emission signals attenuated and modulated through complex paths, structural vibration signals caused by cavitation excitation or non-cavitation factors, and external mechanical noise propagating through the structure. These signals overlap in both the time and frequency domains, especially in the weak state of the initial cavitation phase, where their characteristics are easily masked by other background noise and structural vibrations. Traditional signal processing methods, such as bandpass filtering or amplitude thresholding, struggle to separate these mixed signals and cannot extract identifying features that characterize the degree and type of cavitation. For example, cavitation signals may be submerged by structural resonance peaks, leading to missed detections; or transient structural vibrations caused by non-cavitation may have characteristics similar to a certain cavitation signal, potentially leading to false alarms. This complex signal situation makes the stable identification of early cavitation phenomena in pure water hydraulic systems a technical challenge.

[0008] In the operating environment of pure water hydraulic pumps and valves, the high-frequency acoustic emission signals generated by the collapse of cavitation bubbles undergo reflection, refraction, and attenuation as they propagate through the pure water medium and the complex internal structure of the pumps and valves, resulting in uneven signal energy distribution and characteristic distortion. These signals then cross the water-metal interface and enter the pump and valve housing, coupling with the housing's inherent vibration modes. This results in the received signal being a superposition of the cavitation acoustic emission signal and the pump and valve structural vibration response. Simultaneously, mechanical vibrations generated within the pump and valve and external mechanical noise propagating through the structure are also transmitted to the sensor via the same path. Existing detection methods based on acoustic emission signals struggle to separate and identify this multi-source, cascaded cavitation signal, which is affected by propagation paths and structural coupling. When the cavitation signal is submerged by structural vibration or external noise, it may lead to missed cavitation faults; when non-cavitation-induced transient structural vibrations have similar characteristics to cavitation signals, it may lead to false alarms. This complex and impure nature of the signal makes it difficult to extract identifying features characterizing cavitation based solely on traditional amplitude, energy, or frequency thresholds, hindering early warning and severity assessment of cavitation damage in pure water hydraulic pumps and valves. Summary of the Invention

[0009] The purpose of this invention is to provide a vibration monitoring method and system for pure water hydraulic pump valves, which aims to solve the technical problem of complex superposition of cavitation signals in the operating environment of pure water hydraulic pump valves, which is difficult to separate and identify. Through a multi-sensor collaborative strategy, it can achieve accurate and robust identification of early weak cavitation phenomena, so as to avoid missed and false alarms, thereby effectively warning and assessing cavitation damage.

[0010] In a first aspect, the present invention provides a method for monitoring the vibration of a pure water hydraulic pump valve, comprising the following steps:

[0011] High-frequency acoustic emission signals and low-frequency vibration signals are acquired; both high-frequency acoustic emission signals and low-frequency vibration signals are synchronously acquired on the valve housing of the pure water hydraulic pump.

[0012] The high-frequency acoustic emission signal is processed to obtain the high-frequency characteristic signal;

[0013] The low-frequency vibration signal is processed to obtain the low-frequency characteristic signal;

[0014] Based on high-frequency characteristic signals, identify high-frequency transient events and record the time points and transient characteristics of these events.

[0015] Based on the time point of the high-frequency transient event, extract the low-frequency signal segment that is synchronized with the high-frequency transient event from the low-frequency characteristic signal, and determine the impact vibration characteristics of the low-frequency signal segment.

[0016] The cavitation state of the pure water hydraulic pump valve is determined based on the transient characteristics of high-frequency transient events and the impact vibration characteristics of low-frequency signal bands.

[0017] The vibration monitoring method for pure water hydraulic pump valves provided by this invention achieves coordinated acquisition of cavitation events and their induced structural responses by simultaneously arranging high-frequency acoustic emission sensors and low-frequency vibration sensors on the pure water hydraulic pump valve housing. Utilizing the physical correlation between the high-frequency acoustic emission pulses generated by the collapse of cavitation bubbles and the concurrent impact vibration of the pump valve structure, the unique signal components of cavitation are separated. By extracting features and fusing the separated signals, the cavitation state of the pure water hydraulic pump valve is identified, providing early warning.

[0018] Secondly, the present invention provides a vibration monitoring system for a pure water hydraulic pump valve, comprising:

[0019] The acquisition module is used to acquire high-frequency acoustic emission signals and low-frequency vibration signals; both high-frequency acoustic emission signals and low-frequency vibration signals are synchronously acquired from the pump valve housing of the pure water hydraulic pump.

[0020] The first processing module is used to process the high-frequency acoustic emission signal to obtain the high-frequency characteristic signal;

[0021] The second processing module is used to process the low-frequency vibration signal to obtain the low-frequency characteristic signal;

[0022] The identification and recording module is used to identify high-frequency transient events based on high-frequency characteristic signals, and to record the time points and transient characteristics of the high-frequency transient events;

[0023] The extraction module is used to extract the low-frequency signal segment that is synchronized with the high-frequency transient event from the low-frequency characteristic signal based on the time point of the high-frequency transient event, and to determine the impact vibration characteristics of the low-frequency signal segment.

[0024] The determination module is used to determine the cavitation state of the pure water hydraulic pump valve based on the transient characteristics of high-frequency transient events and the impact vibration characteristics of low-frequency signal bands.

[0025] As can be seen from the above, the vibration monitoring method for pure water hydraulic pump valves provided by this invention effectively distinguishes the signal caused by cavitation from the mechanical vibration of normal pump valve operation and external environmental noise by utilizing the physical correlation between high-frequency acoustic emission signals and low-frequency impact vibrations, significantly reducing the identification difficulty caused by signal aliasing. Furthermore, even if the cavitation acoustic emission signal attenuates and distorts during propagation in pure water media and complex structures, the impact structural vibration response excited on the pump valve housing still retains the characteristics of cavitation impact. This dual confirmation mechanism improves the accuracy of cavitation event identification and reduces the impact of single signal distortion. Finally, this solution requires the cavitation signal to exhibit specific patterns and correlations in both acoustic emission and vibration sensor dimensions, improving the reliability of the judgment. Cavitation is only judged when a specific pulse appears in the high-frequency acoustic emission signal and the synchronous low-frequency vibration signal also exhibits the impact characteristics unique to cavitation, thereby significantly reducing the probability of misjudgment and missed judgment due to complex signal conditions, and achieving early warning and severity assessment of cavitation damage in pure water hydraulic pump valves.

[0026] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing embodiments of the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description and the accompanying drawings. Attached Figure Description

[0027] Figure 1 This is a flowchart of a method for monitoring the vibration of a pure water hydraulic pump valve, provided in an embodiment of the present invention.

[0028] Figure 2 This is a schematic diagram of a pure water hydraulic pump valve vibration monitoring system provided in an embodiment of the present invention.

[0029] Label Explanation:

[0030] 100. Acquisition Module; 200. First Processing Module; 300. Second Processing Module; 400. Identification Record Module; 500. Extraction Module; 600. Confirmation Module. Detailed Implementation

[0031] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0032] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this invention, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0033] Reference Appendix Figure 1 This invention provides a method for monitoring the vibration of a pure water hydraulic pump valve, comprising the following steps:

[0034] High-frequency acoustic emission signals and low-frequency vibration signals are acquired; both high-frequency acoustic emission signals and low-frequency vibration signals are synchronously acquired on the valve housing of the pure water hydraulic pump.

[0035] The high-frequency acoustic emission signal is processed to obtain the high-frequency characteristic signal;

[0036] The low-frequency vibration signal is processed to obtain the low-frequency characteristic signal;

[0037] Based on high-frequency characteristic signals, identify high-frequency transient events and record the time points and transient characteristics of these events.

[0038] Based on the time point of the high-frequency transient event, extract the low-frequency signal segment that is synchronized with the high-frequency transient event from the low-frequency characteristic signal, and determine the impact vibration characteristics of the low-frequency signal segment.

[0039] The cavitation state of the pure water hydraulic pump valve is determined based on the transient characteristics of high-frequency transient events and the impact vibration characteristics of low-frequency signal bands.

[0040] High-frequency acoustic emission signals refer to the high-frequency elastic wave signals generated by the collapse of cavitation bubbles, typically ranging from tens of kilohertz to several megahertz. Low-frequency vibration signals refer to the low-frequency vibration response signals excited by cavitation impacts or other mechanical movements on the pump / valve structure, typically ranging from hundreds of hertz to thousands of hertz. Both signals are synchronously acquired on the pump / valve housing of the pure water hydraulic pump. Synchronous acquisition means using different types of sensors, such as piezoelectric acoustic emission sensors and accelerometers, to simultaneously acquire high-frequency acoustic emission signals and low-frequency vibration signals at the same time reference. The purpose is to ensure the temporal correlation of the two signals, providing a time alignment basis for subsequent joint analysis. High-frequency characteristic signals refer to signals that highlight the characteristics of cavitation transient events after preprocessing the original high-frequency acoustic emission signals. These signals can be achieved using techniques such as filtering, noise reduction, and envelope extraction. The purpose is to remove noise interference, enhance the signal-to-noise ratio of the cavitation signal, and facilitate subsequent transient event identification. Low-frequency characteristic signals refer to signals that reflect the structural impact response characteristics obtained after preprocessing the original low-frequency vibration signals. These signals can be achieved using techniques such as filtering, noise reduction, and demodulation. The aim is to remove irrelevant vibration components and highlight the low-frequency response related to cavitation impact, facilitating subsequent synchronous extraction and feature analysis. Identifying high-frequency transient events involves determining multiple candidate transient events from the high-frequency characteristic signals based on changes in signal energy or amplitude. Time-frequency analysis and waveform morphology analysis are then used to determine whether these events conform to preset cavitation event characteristics, aiming to capture the moment cavitation occurs. Recording the time point and transient characteristics of high-frequency transient events involves recording the time of occurrence and instantaneous characteristics, such as instantaneous amplitude, instantaneous energy, pulse rate, and specific frequency band energy, after identifying the high-frequency transient event. The purpose is to quantify the intensity and characteristics of the cavitation event, serving as a direct basis for judging the cavitation state. Extracting low-frequency signal segments synchronized with high-frequency transient events refers to acquiring low-frequency vibration signal segments that are time-aligned with the high-frequency event from low-frequency characteristic signals based on the time point of the high-frequency transient event. This can be achieved using methods such as cross-correlation analysis or synchronous time-domain superposition. The purpose is to obtain the synchronous low-frequency response of cavitation impact on the pump and valve structure. Determining the impact vibration characteristics of the low-frequency signal segments involves performing impact vibration characteristic analysis and aperiodic analysis on the extracted low-frequency signal segments to quantify their impact vibration characteristics, such as impact energy, peak factor, kurtosis, and pulse duration. The purpose is to quantify the degree of impact of cavitation impact on the structure, serving as a supplementary basis for judging the cavitation state. Determining the cavitation state of the pure water hydraulic pump valve involves a comprehensive analysis based on the transient characteristics of the high-frequency transient event and the impact vibration characteristics of the low-frequency signal segments to determine whether cavitation exists in the pump valve, the degree of cavitation, and the possible types of cavitation. The purpose is to achieve early warning and assessment of cavitation damage to the pure water hydraulic pump valve.

[0041] The working principle of this invention lies in the fact that the collapse of cavitation bubbles is an instantaneous high-energy release process. It not only generates high-frequency acoustic emission waves, but these high-frequency waves also excite impact-induced low-frequency structural vibrations when impacting the pump and valve structure. There is a unique physical correlation between this "high-frequency acoustic emission event" and the "synchronous low-frequency impact vibration," which is a unique "fingerprint" of the cavitation phenomenon. This solution analyzes the pump and valve structure response captured by a low-frequency vibration sensor within the same time window when the high-frequency acoustic emission sensor detects the direct signal (high-frequency pulse) of the cavitation event. If the low-frequency vibration signal also exhibits non-periodic impact characteristics synchronized with the high-frequency acoustic emission event (e.g., a significant increase in kurtosis), the event can be highly confidently attributed to cavitation. Conversely, if a high-frequency signal appears but the low-frequency vibration has no corresponding impact characteristics, or if the low-frequency vibration is periodic rather than impact-induced, cavitation can be ruled out. Through this dual verification and feature fusion, the cavitation signal is effectively distinguished from the mechanical vibrations of normal pump and valve operation, external environmental noise, and other non-cavitation-induced transient structural vibrations, thereby achieving accurate identification and early warning of cavitation phenomena.

[0042] The core innovation of this application lies in the synchronous acquisition and joint analysis of high-frequency acoustic emission signals and low-frequency vibration signals. The high-frequency signals are used to identify cavitation transient events and obtain their transient characteristics. Then, using these as time anchors, synchronous impact vibration characteristics are extracted from the low-frequency signals. This allows for a comprehensive judgment of the cavitation state of the pure water hydraulic pump valve, solving the problem that a single signal source is difficult to identify and assess the cavitation state in complex media, and improving the accuracy and robustness of cavitation monitoring.

[0043] Specifically, this method monitors the cavitation state of a pure water hydraulic pump valve through a multi-source signal collaborative analysis strategy. First, high-frequency acoustic emission signals and low-frequency vibration signals are simultaneously acquired from the pump valve housing, ensuring the temporal correspondence between the two signals and laying the foundation for subsequent joint analysis, thereby capturing information about cavitation phenomena across different frequency ranges. Then, these two raw signals are processed separately to remove noise and highlight their respective characteristics, yielding high-frequency and low-frequency characteristic signals, providing input for subsequent identification and extraction. Next, utilizing the transient characteristics of the high-frequency characteristic signals, high-frequency transient events generated by the collapse of cavitation bubbles are identified, and the time points and transient characteristics of these events are recorded. High-frequency transient events, as direct representations of cavitation phenomena, serve as anchor points connecting high-frequency and low-frequency signals. Based on this, using the time points of the high-frequency transient events as a benchmark, low-frequency signal segments synchronized with the high-frequency events are extracted from the low-frequency characteristic signals. This synchronous extraction mechanism based on the time points of high-frequency events ensures that the analyzed low-frequency signal segments are the response of cavitation impacts to the pump valve structure, rather than other unrelated mechanical vibrations. Simultaneously, impact vibration characteristic analysis was performed on this low-frequency signal band to quantify the impact of cavitation impact on the structure. Finally, by integrating the transient characteristics of high-frequency transient events and the impact vibration characteristics of the low-frequency signal band, a comprehensive judgment was made to determine the cavitation state of the pure water hydraulic pump valve. This complementary analysis of high and low frequency signals distinguishes cavitation signals from background noise and non-cavitation vibrations, improving the accuracy and reliability of cavitation state determination.

[0044] As one embodiment, the solution of this application is specifically implemented as follows: A piezoelectric acoustic emission sensor and an accelerometer are respectively installed on the outer surface of the housing of the pure water hydraulic pump valve, such as on the pump body or valve body, and connected to a multi-channel data acquisition system to achieve synchronous acquisition of high-frequency acoustic emission signals and low-frequency vibration signals. The acquired raw signals are transmitted to a data processing unit via a data line, which can be a computer or an embedded controller. In this processing unit, the high-frequency acoustic emission signal is first filtered by a bandpass filter to remove low-frequency mechanical noise and high-frequency environmental noise, and envelope demodulation is performed to obtain high-frequency characteristic signals. At the same time, the low-frequency vibration signal is also obtained by bandpass filtering and time-domain analysis. Subsequently, the processing unit processes the high-frequency characteristic signals using a transient event detection algorithm, such as pulse recognition based on short-time energy or amplitude thresholds, to identify high-frequency transient events and record the occurrence time of each event and its transient characteristics such as instantaneous energy and pulse width. Based on the occurrence times of these high-frequency transient events, the processing unit delineates time windows in the low-frequency characteristic signals and performs cross-correlation analysis or synchronous time-domain superposition to extract low-frequency signal segments synchronized with the high-frequency transient events. For the extracted low-frequency signal segments, impact indices such as kurtosis, peak factor, and impact energy are calculated to determine their impact vibration characteristics. Finally, the processing unit inputs the transient characteristics of the high-frequency transient events and the impact vibration characteristics of the low-frequency signal segments into a trained classification model, such as a support vector machine or neural network, or performs logical judgment based on an expert rule base to determine the current cavitation state of the pure water hydraulic pump valve, such as no cavitation, slight cavitation, or severe cavitation.

[0045] By employing the aforementioned solution, this application addresses the problem of cavitation acoustic emission signals being superimposed with structural vibrations and external noise during the operation of pure water hydraulic pumps and valves. Furthermore, the signals undergo distortion during propagation in complex media and structures, making it difficult to identify and assess the cavitation state. This method achieves multi-dimensional monitoring of cavitation phenomena by simultaneously acquiring and jointly analyzing high-frequency acoustic emission signals and low-frequency vibration signals. High-frequency signals are used to identify the occurrence of cavitation transient events, while low-frequency signals are used to quantify the structural response to cavitation impacts. This combination distinguishes cavitation phenomena from background noise or vibrations caused by non-cavitation. Therefore, this method improves the accuracy and robustness of cavitation state assessment, enabling early warning and severity assessment of cavitation damage to pure water hydraulic pumps and valves. This avoids equipment performance degradation and failure due to cavitation erosion, ensuring the operation of the pure water hydraulic system.

[0046] In some embodiments, the transient characteristics of a high-frequency transient event include instantaneous amplitude, instantaneous energy, pulse rate, and energy in a specific frequency band.

[0047] In some embodiments, the step of identifying high-frequency transient events based on high-frequency characteristic signals includes:

[0048] From high-frequency characteristic signals, multiple candidate transient events are identified based on changes in signal energy or amplitude, and the start time and duration of each candidate transient event are recorded.

[0049] By performing time-frequency analysis on each candidate transient event, the time-frequency characteristics of each candidate transient event are obtained; the time-frequency characteristics include the energy distribution and duration of the candidate transient event within a specific frequency range;

[0050] By performing waveform morphology analysis on each candidate transient event, the waveform morphology characteristics of each candidate transient event are obtained; the waveform morphology characteristics include the pulse width, rise steepness, and attenuation characteristics of the candidate transient events.

[0051] Based on time-frequency characteristics and waveform morphology features, high-frequency transient events are identified by judging whether each candidate transient event conforms to the preset cavitation event characteristics.

[0052] Time-frequency analysis (TF-F) is a signal processing technique used to analyze how the frequency components of a signal change over time. It reveals the spectral content of a signal at different points in time, providing more comprehensive information than time-domain or frequency-domain analysis alone. TF-F analysis allows us to obtain the energy distribution and duration of a signal within a specific frequency range, which is crucial for identifying cavitation signals with transient and broadband characteristics. Waveform morphology analysis is a method for quantitatively describing the shape of a signal waveform. It focuses on the geometric characteristics of the signal, such as pulse width, the speed at which the signal rises from the baseline to the peak, and the speed and manner in which the signal falls from the peak to the baseline. These characteristics reflect the physical generation mechanism of the signal and are important for distinguishing transient events from different sources. Predefined cavitation event features refer to a set of criteria established based on the physical understanding of cavitation phenomena, accumulated experimental data, and expert experience. These criteria define the typical manifestations of cavitation events in the time-frequency domain and waveform morphology, and can exist in the form of rule sets, threshold ranges, or machine learning model parameters, used for classifying candidate transient events.

[0053] This application addresses the problem of cavitation signals being easily distorted due to complex propagation paths and structural coupling during the operation of pure water hydraulic pumps and valves. It proposes a multi-dimensional feature fusion-based transient event identification method. This method aims to accurately identify high-frequency transient cavitation events, thus providing reliable input for subsequent cavitation state determination. Specifically, the method first preliminarily filters out all potential transient impacts from continuous high-frequency characteristic signals by monitoring significant changes in signal energy or amplitude, identifying them as candidate transient events. This preliminary screening step can broadly capture all transient energy releases in the signal, ensuring that no possible cavitation events are missed, and records the start time and duration of each captured candidate event, laying the foundation for subsequent refined analysis. Building on this, to overcome the identification challenges posed by signal distortion, this method further performs time-frequency analysis on each candidate transient event. Through time-frequency analysis, the variation law of energy distribution of the signal at different frequencies over time can be revealed, thereby obtaining the time-frequency characteristics of the candidate transient event, such as energy distribution and duration within a specific frequency range. The acoustic emission signal generated by the collapse of cavitation bubbles has its inherent time-frequency fingerprint, such as energy concentrated in a specific high-frequency band and extremely short duration. This analysis, capable of deeply exploring the frequency dynamics of a signal, allows for the effective capture of the energy distribution pattern in the time-frequency domain even when the original cavitation signal waveform is distorted. This helps distinguish genuine cavitation events from background noise or transient vibrations caused by non-cavitation. Simultaneously, this method performs morphological analysis on the waveforms of each candidate transient event to obtain their waveform morphological characteristics, including pulse width, rise steepness, and attenuation characteristics. The shock waves generated by cavitation events have specific waveform geometric characteristics; for example, their pulses are typically narrow, with a steep rise and rapid attenuation. Because these morphological characteristics reflect the physical essence of cavitation events and differ significantly from the waveforms of other transient events such as mechanical shocks or structural resonances, the analysis of these waveform details allows for the extraction of the "fingerprint" information of cavitation events. This provides richer criteria for subsequent accurate identification and effectively eliminates interference that may have some similarities to the cavitation signal in the time-frequency domain but whose waveform morphology does not match. Finally, this method comprehensively judges the characteristics obtained from time-frequency analysis and the features obtained from waveform morphological analysis. By comparing these multi-dimensional features with preset cavitation event features, only candidate events that simultaneously meet multiple cavitation-specific discrimination criteria are confirmed as high-frequency transient cavitation events. This multi-dimensional, multi-feature fusion judgment mechanism avoids false alarms and missed alarms that may be caused by judging a single feature. As a key part of the entire pure water hydraulic pump valve vibration monitoring method, this identification process significantly enhances the entire monitoring system's ability to perceive early and weak cavitation phenomena by providing more accurate and robust high-frequency transient event identification results.It is precisely because these high-frequency transient events can be identified stably and accurately that the subsequent steps of extracting synchronous low-frequency signal segments from low-frequency characteristic signals and determining impact vibration characteristics can be based on more reliable event triggering, thereby ultimately achieving accurate determination of the cavitation state of pure water hydraulic pump valves, effectively improving the reliability and early warning capability of the entire monitoring method.

[0054] As a preferred implementation, the process of identifying high-frequency transient events based on high-frequency characteristic signals can be carried out as follows: First, when determining multiple candidate transient events from the high-frequency characteristic signals, a peak detection algorithm based on an adaptive threshold can be used. For example, the root mean square value of the short-term energy or instantaneous amplitude of the high-frequency characteristic signal can be calculated, and a dynamic threshold can be set, which can be adjusted in real time according to the background noise level of the signal. When the energy or amplitude of the signal exceeds the threshold and persists for a period of time, it is identified as a candidate transient event. Simultaneously, the time point at which the event first exceeds the threshold can be recorded as the starting time point, and the time from the starting point to the energy or amplitude falling back below the threshold can be recorded as the duration. Further, when performing time-frequency analysis on each candidate transient event, wavelet transform or short-time Fourier transform can be used. For example, continuous wavelet transform can be performed on the signal segment of each candidate transient event to generate its time-frequency graph. From this time-frequency graph, the energy concentration within a preset cavitation characteristic frequency range can be extracted, and the distribution width of this energy on the time axis can be calculated, thereby obtaining the energy distribution and duration of the candidate transient event within a specific frequency range. Furthermore, when performing morphological analysis on the waveforms of each candidate transient event, its pulse width, rise edge steepness, and attenuation characteristics can be extracted. For example, the peak point of the candidate transient event waveform can be identified, and its half-peak width can be calculated as the pulse width. The rise edge steepness can be obtained by calculating the ratio of the time required for the waveform to rise from its baseline value to 90% of its peak value to the amplitude change. The attenuation characteristics can be quantified by performing an exponential fit on the waveform's falling segment and extracting its attenuation constant. Finally, when determining whether each candidate transient event meets the preset cavitation event characteristics based on its time-frequency characteristics and waveform morphological features, a multi-feature classifier can be constructed. This classifier can be a rule-based expert system, in which thresholds or ranges for the time-frequency characteristics and waveform morphological features of cavitation events are preset. When the time-frequency characteristics and waveform morphological features of a candidate transient event simultaneously meet these preset conditions, the event is identified as a high-frequency transient cavitation event. As another approach, machine learning models, such as support vector machines or neural networks, can be used. By training on a large number of known samples of cavitation and non-cavitation events, the model can automatically learn and identify patterns that match the characteristics of cavitation events. In this way, accurate identification of high-frequency transient events can be achieved.

[0055] This method effectively addresses the problem of reduced accuracy in traditional single-feature identification methods due to distortion of cavitation signals in pure water hydraulic pump valves under complex propagation paths by performing multi-dimensional analysis of high-frequency characteristic signals. By comprehensively considering the time-frequency characteristics and waveform morphology of candidate transient events, this method can more accurately capture the inherent physical fingerprint of cavitation events, effectively distinguishing them from background noise or transient vibrations caused by non-cavitation even when the signal is interfered with or distorted. This significantly improves the accuracy and robustness of identifying early, weak cavitation phenomena in complex operating environments, effectively avoiding false alarms and missed alarms of cavitation events, and providing reliable technical support for early fault warning of pure water hydraulic pump valves.

[0056] In some embodiments, the step of extracting a low-frequency signal segment synchronized with the high-frequency transient event from the low-frequency characteristic signal based on the time point of the high-frequency transient event includes:

[0057] Based on the transient characteristics of high-frequency transient events, the propagation response characteristics of high-frequency transient events in pure water medium and pump valve structure are obtained by analyzing the energy decay characteristics and duration of high-frequency transient events.

[0058] The initial time window is determined based on the propagation response characteristics. The initial time window is centered on the time point of the high-frequency transient event, and its length is adjusted according to the duration of the propagation response characteristics.

[0059] Within the initial time window, the first analysis result is obtained by performing cross-correlation analysis on the envelope signal and low-frequency characteristic signal of the high-frequency transient event;

[0060] Based on the results of the first analysis, a precisely aligned time window is formed by determining the start and end times of the low-frequency signal band.

[0061] Low-frequency signal segments are extracted from low-frequency characteristic signals based on precisely aligned time windows.

[0062] Propagation response characteristics refer to the comprehensive manifestation of the changes in energy, waveform, time delay, and duration of a high-frequency transient event propagating in pure water media and pump / valve structures, depending on the propagation path and media characteristics. These characteristics can be obtained using signal attenuation models, multipath propagation models, or structural vibration transfer function analysis. The initial time window refers to a pre-defined time range used to define the low-frequency response signal that the high-frequency transient event may excite. Its length and center point can be determined using empirical values, preset models, or real-time signal characteristic analysis. The envelope signal of a high-frequency transient event refers to the trajectory of the instantaneous amplitude or energy of the event signal over time, reflecting the fluctuations and persistence of the signal energy. It can be extracted using Hilbert transform, rectified filtering, or squared averaging. Cross-correlation analysis is a signal processing technique used to measure the similarity between two signals at different time delays. It can be implemented using time-domain cross-correlation operations, frequency-domain cross-correlation operations, or wavelet transform-based cross-correlation analysis. The first analysis result refers to the information obtained after cross-correlation analysis regarding the time synchronization relationship and correlation strength between the high-frequency transient event envelope signal and the low-frequency characteristic signal. This information can be expressed as the peak value, peak position, or correlation coefficient of the cross-correlation function. The precisely aligned time window refers to the time range formed after accurately locating the start and end times of the low-frequency signal segment based on the cross-correlation analysis results. This window can be determined using methods such as cross-correlation peak position, signal energy threshold, or signal morphology characteristics.

[0063] This solution addresses the challenge of accurately determining the low-frequency response time window due to the complex propagation characteristics of cavitation signals in pure water hydraulic pumps and valves. It proposes a method for precise extraction of low-frequency signal segments based on the characteristics of high-frequency transient events and signal cross-correlation analysis. The core principle lies in the inherent physical correlation between the high-frequency acoustic emission events generated by the collapse of cavitation bubbles and the synchronously excited low-frequency structural vibrations. This correlation is reflected in the propagation response characteristics of the high-frequency events and the temporal correlation between the high-frequency signal envelope and the low-frequency signal. This solution first utilizes the transient characteristics of high-frequency transient events, analyzing their energy attenuation characteristics and duration to obtain the propagation response characteristics of high-frequency transient events in pure water media and pump / valve structures. This step is crucial for understanding how high-frequency cavitation signals are transformed into low-frequency structural responses. By deeply analyzing the energy attenuation and duration characteristics of high-frequency transient events themselves, it is possible to infer their potential time delays, energy dispersion, and the duration of structural excitation during propagation in pure water media and complex pump / valve structures. This provides important prior information for subsequently determining the possible range of low-frequency signal segments, overcoming the limitations of extraction based solely on a single time point. Based on this, an initial time window is determined according to the acquired propagation response characteristics. This initial time window is based on the occurrence time of the high-frequency transient event and its length is dynamically adjusted according to the signal duration or possible delay range revealed by the propagation response characteristics. This ensures that the initial time window is sufficiently broad to initially cover the possible time range of low-frequency impact vibrations excited by the high-frequency transient event, avoiding the problem of missing key low-frequency responses due to an overly narrow time window. Furthermore, within this initial time window, the first analysis result is obtained by performing cross-correlation analysis on the envelope signal of the high-frequency transient event and the low-frequency characteristic signal. This is the core step in achieving precise alignment. The envelope signal of the high-frequency transient event reflects its energy change trend over time, while the low-frequency characteristic signal contains the structural impact vibrations excited by the high-frequency event. It is by performing cross-correlation analysis on these two that the strongest correlation between them can be effectively identified, that is, the precise temporal correspondence between the low-frequency impact vibrations and the high-frequency transient event. Cross-correlation analysis effectively suppresses noise interference and identifies time delays between signals, yielding a reliable initial analysis result that lays the foundation for subsequent precise time window determination. Subsequently, based on the initial analysis result, the start and end times of the low-frequency signal segment can be precisely located, forming a more accurate and aligned time window than the initial one. This precisely aligned time window can more accurately define the low-frequency impact vibrations caused by high-frequency transient events, eliminating interference from irrelevant signals. Finally, based on this precisely aligned time window, the low-frequency signal segment is extracted from the low-frequency characteristic signal. Through this series of refined processes, a low-frequency signal segment precisely synchronized and aligned with the high-frequency transient event is ultimately obtained.The extracted low-frequency signal segment retains the impact vibration characteristics caused by cavitation events to the greatest extent, providing high-quality input for the subsequent accurate determination of the impact vibration characteristics of the low-frequency signal segment. Overall, this scheme achieves accurate extraction of the low-frequency signal segment by dynamically analyzing the propagation response characteristics of high-frequency transient events and combining the cross-correlation analysis of the high-frequency envelope signal and the low-frequency characteristic signal. This overcomes the non-constant nature of signal propagation in the complex environment of pure water hydraulic pumps and valves, ensuring that the extracted low-frequency signal segment is highly synchronized with the high-frequency transient event in time. It is this precise synchronous extraction that makes the subsequent determination of the impact vibration characteristics of the low-frequency signal segment more accurate, thereby significantly improving the accuracy and robustness of identifying early weak cavitation phenomena and providing more reliable low-frequency dimension evidence for judging the cavitation state of pure water hydraulic pumps and valves.

[0064] In one specific implementation, to extract a low-frequency signal segment synchronized with a high-frequency transient event from a low-frequency characteristic signal based on the time point of the high-frequency transient event, the energy decay characteristics and duration can first be analyzed by performing time-frequency analysis on the high-frequency transient event, such as its instantaneous amplitude, instantaneous energy, or specific frequency band energy, using methods like short-time Fourier transform or wavelet transform. By observing the energy peak of the high-frequency signal within a specific frequency range and the time required for it to decrease from the peak to the background noise level, the propagation response characteristics of the high-frequency transient event in pure water media and pump / valve structures can be obtained, such as the possible time delay and energy dispersion during propagation within the structure. Based on this, an initial time window can be determined according to the obtained propagation response characteristics. This initial time window can be centered on the start time point of the high-frequency transient event and its length can be set according to the maximum possible time delay and typical duration of the low-frequency response determined in the propagation response characteristics. For example, if the propagation response characteristics indicate that the low-frequency response may last for 15 milliseconds, the initial time window can be set to 10 milliseconds before and after the high-frequency event time point to ensure complete coverage of the response. Subsequently, within this initial time window, cross-correlation analysis can be performed on the envelope signal of the high-frequency transient event and the low-frequency characteristic signal. The envelope signal of the high-frequency transient event can be obtained by performing a Hilbert transform on its original signal and taking the modulus. The cross-correlation analysis can be performed using a Fast Fourier Transform (FFT) to perform frequency domain cross-correlation operations to obtain a cross-correlation function. The maximum peak value in this cross-correlation function and its corresponding time offset can be used as the first analysis result. Then, based on this first analysis result, i.e., the time offset of the cross-correlation peak value, the start and end time points of the low-frequency signal segment can be accurately located. For example, the time point corresponding to the cross-correlation peak value can be used as a reference to trace back a preset time (e.g., 2 milliseconds) as the starting point, and extend backward until the energy of the low-frequency signal drops to the background noise level (e.g., 10% of the peak energy) as the ending point, thus forming a precisely aligned time window. Finally, based on this precisely aligned time window, the low-frequency signal segment can be extracted from the low-frequency characteristic signal. The extracted low-frequency signal segment retains the impact vibration characteristics caused by cavitation events to the greatest extent, providing high-quality input for the subsequent accurate determination of the impact vibration characteristics of the low-frequency signal segment.

[0065] This scheme achieves accurate extraction of low-frequency signal segments by analyzing the propagation response characteristics of high-frequency transient events and combining the cross-correlation analysis of the envelope signal of the high-frequency transient event with low-frequency characteristic signals. This overcomes the non-fixed time delay and energy attenuation characteristics of cavitation signal propagation in the complex environment of pure water hydraulic pumps and valves, and avoids the problems of missed low-frequency impact responses or background noise introduction that may occur when using a fixed time window. By forming a precisely aligned time window, it ensures that the extracted low-frequency signal segments are highly synchronized with the high-frequency transient events in time, thereby improving the accuracy of synchronization judgment and providing accurate and reliable low-frequency dimension evidence for subsequent cavitation state judgment, thus improving the accuracy and robustness of identifying early weak cavitation phenomena.

[0066] In some embodiments, the step of obtaining a first analysis result by performing cross-correlation analysis on the envelope signal of the high-frequency transient event and the low-frequency characteristic signal within an initial time window includes:

[0067] Within the initial time window, cross-correlation is performed on the envelope signal of the high-frequency transient event and the low-frequency characteristic signal to obtain the cross-correlation function;

[0068] Identify multiple cross-correlation peaks from the cross-correlation function, and obtain the time offset and peak intensity of each cross-correlation peak;

[0069] By performing morphological analysis on each cross-correlation peak, the morphological characteristics of each cross-correlation peak are obtained; the morphological characteristics include the width, symmetry, and tailing characteristics of the cross-correlation peak.

[0070] By combining the transient characteristics of high-frequency transient events, and by comprehensively judging the time offset, peak intensity and morphological characteristics of cross-correlation peaks, cross-correlation peaks that conform to the unique correlation pattern of cavitation events can be identified.

[0071] Based on the identified cross-correlation peaks, the precise time delay information between the high-frequency transient event and the low-frequency characteristic signal is determined and used as the first analysis result.

[0072] Cross-correlation peaks refer to local maxima in the cross-correlation function, indicating a high degree of similarity or correlation between two signals at a specific time offset. Morphological analysis involves structurally examining and quantifying the waveform of cross-correlation peaks to extract their geometric or shape features. This can be achieved using mathematical fitting, feature point extraction, or pattern recognition algorithms. Morphological features describe the shape of cross-correlation peaks, such as peak width, symmetry, and tailing characteristics. These features reflect the intrinsic physical processes or propagation characteristics of signal correlation. Cavitation event-specific correlation patterns refer to the specific and distinguishable correlation patterns exhibited on the cross-correlation function between the high-frequency acoustic emission signal generated by cavitation bubble collapse and the synchronously excited low-frequency structural vibrations. These patterns are reflected not only in the time offset and peak intensity but also in the specific morphological characteristics of the cross-correlation peaks. Comprehensive judgment refers to a unified evaluation and decision-making process that combines data or features from multiple dimensions. This can involve methods such as weight allocation, logical rules, machine learning algorithms, or expert systems to improve the accuracy and robustness of the judgment.

[0073] In the complex environment of cavitation detection in pure water hydraulic pump valves, this paper proposes a refined discrimination method for cross-correlation results based on multi-dimensional feature fusion to address the misjudgment caused by non-cavitation transient events in cross-correlation analysis. This method first performs cross-correlation calculations on the envelope signal of the high-frequency transient event and the low-frequency characteristic signal within an initial time window, thus obtaining the cross-correlation function. This basic operation quantifies the similarity between the two signals at different time offsets, initially revealing the possible correlation between the high-frequency transient event and the low-frequency characteristic signal. Since multiple interference sources exist in actual operation, multiple peaks may appear in the cross-correlation function. These peaks may correspond to the propagation delay of the actual cavitation event or be caused by non-cavitation factors. Therefore, this scheme further identifies multiple cross-correlation peaks from the cross-correlation function and obtains the time offset and peak intensity of each peak, providing candidate targets for subsequent refined screening.

[0074] To distinguish between genuine cavitation correlations and spurious correlations caused by interference, this method performs morphological analysis on each cross-correlation peak to obtain its morphological characteristics, such as peak width, symmetry, and tailing properties. The correlation patterns of signals generated by cavitation bubble collapse exhibit specific morphological fingerprints during propagation and coupling, while spurious peaks caused by noise or non-cavitation vibrations may display different morphologies. Analyzing these morphological characteristics can provide additional criteria for distinguishing genuine cavitation correlations.

[0075] Furthermore, this scheme combines the transient characteristics of acquired high-frequency transient events, such as instantaneous amplitude, instantaneous energy, pulse rate, and specific frequency band energy, to comprehensively judge the cross-correlation peak's time offset, peak intensity, and morphological characteristics. This multi-dimensional information fusion mechanism can construct a more comprehensive judgment model. If a high-frequency transient event has typical cavitation transient characteristics, then its corresponding low-frequency signal segment should also exhibit peak morphology and intensity consistent with cavitation propagation characteristics in the cross-correlation function. This comprehensive judgment mechanism significantly improves the accuracy of identifying cavitation event correlations and effectively reduces the risk of false alarms and missed alarms.

[0076] Finally, based on the identified cross-correlation peaks conforming to the unique correlation pattern of cavitation events, the precise time delay information between the high-frequency transient event and the low-frequency characteristic signal is determined and used as the first analysis result. This precise time delay information effectively avoids interference from non-cavitation transient events, ensuring that the low-frequency signal segments subsequently extracted from the low-frequency characteristic signal are truly synchronized with the cavitation event, thus laying a solid foundation for subsequent impact vibration characteristic analysis and cavitation state determination. This scheme overcomes the limitation of traditional cross-correlation analysis in complex pure water hydraulic pump and valve environments, which is susceptible to interference from non-cavitation transient events, by performing multi-dimensional and refined discrimination on the cross-correlation analysis results. This results in higher purity and representativeness of the synchronized signal segments extracted from the low-frequency characteristic signal, thereby improving the overall accuracy and reliability of cavitation detection.

[0077] In one implementation, to perform cross-correlation analysis on the envelope signal of a high-frequency transient event and the low-frequency characteristic signal within an initial time window and obtain a first analysis result, a cross-correlation operation can first be performed on the envelope signal of the high-frequency transient event and the low-frequency characteristic signal to obtain a cross-correlation function. For example, a Fast Fourier Transform (FFT) can be used to transform the two signals to the frequency domain, perform conjugate multiplication, and then use an Inverse Fast Fourier Transform (IFFT) to obtain the time-domain cross-correlation function. Subsequently, multiple cross-correlation peaks are identified from this cross-correlation function, and the time offset and peak intensity of each cross-correlation peak are obtained. This can be achieved by setting a dynamic threshold and combining it with a local maximum search algorithm to ensure that all potential peaks are identified. For each identified peak, its time position in the cross-correlation function is recorded as the time offset, and its corresponding function value is recorded as the peak intensity. Next, morphological analysis can be performed on each cross-correlation peak to obtain the morphological characteristics of each cross-correlation peak. For example, curve fitting, such as Gaussian function or Lorentz function fitting, can be performed on each peak region to quantify the width of the peak. The symmetry of the peak can be assessed by comparing the slope or area on both sides of the peak. Tail characteristics can be determined by analyzing the rate and shape of the peak decay from the apex to both sides, for example, by calculating the time or distance required to decay to a certain proportion. Furthermore, by combining the transient characteristics of high-frequency transient events, such as their instantaneous amplitude, instantaneous energy, impulse rate, and specific frequency band energy, and by comprehensively judging the time offset, peak intensity, and morphological characteristics of the cross-correlation peaks, cross-correlation peaks that conform to the unique correlation patterns of cavitation events can be identified. This allows the construction of a multi-input decision model, for example, training a support vector machine (SVM) or a shallow neural network classifier. The input features of this classifier can include the time offset, peak intensity, width, symmetry, and tail characteristics of the cross-correlation peaks, as well as the instantaneous amplitude, instantaneous energy, impulse rate, and specific frequency band energy of the corresponding high-frequency transient events. By training on a large number of known cavitation and non-cavitation event samples, this model can learn and identify the unique patterns exhibited by cavitation events on the cross-correlation function. Finally, based on the cross-correlation peaks that conform to the unique correlation pattern of cavitation events identified by the model, the precise time delay information between the high-frequency transient event and the low-frequency characteristic signal is determined and used as the first analysis result. For example, if the model identifies a peak as being related to a cavitation event, then the time offset corresponding to that peak is determined as the precise time delay information.

[0078] This scheme effectively distinguishes between true correlations caused by cavitation and false correlations caused by non-cavitation factors by incorporating multi-dimensional features into cross-correlation analysis. Through detailed analysis of the time offset, peak intensity, and morphological characteristics of cross-correlation peaks, combined with the transient characteristics of high-frequency transient events themselves, cross-correlation peaks conforming to the unique correlation patterns of cavitation events can be identified. This significantly reduces the misjudgment rate caused by non-cavitation transient events in the cavitation detection of pure water hydraulic pump valves, improving the accuracy and reliability of cavitation detection. The final determined precise time delay information ensures the synchronization between the subsequently extracted low-frequency signal segment and the actual cavitation event, providing a reliable data foundation for subsequent cavitation state determination.

[0079] In some embodiments, the impact vibration characteristics of the low-frequency signal band are determined according to the following steps:

[0080] The second analysis result was obtained by performing impact vibration characteristic analysis and non-periodic analysis on the low-frequency signal band;

[0081] The energy distribution of the low-frequency signal band is obtained, and the impact vibration characteristics of the low-frequency signal band are determined by combining the second analysis results.

[0082] Shock vibration characteristic analysis refers to the process of identifying and quantifying the instantaneous, high-energy, short-duration shock components in a signal. This can be calculated using statistical indicators such as peak factor, kurtosis, impulse factor, and impact factor, or extracted using time-frequency analysis methods such as wavelet transform and envelope demodulation. Aperiodicity analysis refers to the process of evaluating the intensity and regularity of periodic components in a signal, thereby distinguishing random, transient events from periodic or quasi-periodic events. This can be achieved using autocorrelation function analysis, power spectral density analysis, cyclostationary analysis, or machine learning-based pattern recognition methods to assess the periodicity or aperiodicity of the signal. The second analysis result refers to the comprehensive data or indicator set obtained through shock vibration characteristic analysis and aperiodicity analysis. It can be a multi-dimensional feature vector containing shock intensity indicators, aperiodic indicators, and other relevant statistics, or a single discriminant value after feature fusion. Energy distribution refers to the energy intensity of a low-frequency signal band in the time or frequency dimension and its variation with time or frequency. It can be represented by the root mean square value, signal energy integral, short-time energy, or the sum of energy within a specific frequency band. Combining the results of the second analysis refers to the process of comprehensively considering and fusing the energy distribution of the low-frequency signal band with the results of the second analysis. This can employ various data fusion or pattern recognition algorithms, such as weighted summation, decision trees, support vector machines, and neural networks, to form a more comprehensive basis for judgment. Determining the impact vibration characteristics of the low-frequency signal band refers to deriving quantitative indicators or judgment conclusions based on the comprehensive analysis results that can accurately characterize the impact vibration characteristics caused by cavitation in the low-frequency signal band. This can be a numerical value, a classification result, or a set of feature parameters.

[0083] This scheme aims to accurately determine the impact vibration characteristics of low-frequency signal segments extracted from low-frequency characteristic signals, overcoming the difficulty of accurately identifying cavitation impacts in complex signal aliasing environments using traditional methods. First, by analyzing the impact vibration characteristics of the low-frequency signal segments, the instantaneous, high-energy impact components in the signal can be identified and quantified. This analysis directly targets the typical impact characteristics generated by the collapse of cavitation bubbles, helping to highlight cavitation events from background noise. Simultaneously, non-periodic analysis is a key aspect of this scheme. During operation, the internal mechanical motion of pure water hydraulic pump valves generates periodic or quasi-periodic vibration signals, which may overlap with the low-frequency impact signals caused by cavitation in the frequency domain. Non-periodic analysis can effectively distinguish between random, non-periodic impact events caused by cavitation and periodic vibrations generated by mechanical motion, thereby eliminating the interference of non-cavitation factors on the determination of impact vibration characteristics and obtaining analytical results. Based on this, obtaining the energy distribution of the low-frequency signal segments allows for a comprehensive understanding of the signal's energy intensity and distribution patterns. Energy is an important indicator for measuring the intensity of impact events, and cavitation impacts are usually accompanied by the instantaneous release of energy. More importantly, this scheme combines the obtained energy distribution with the aforementioned analysis results. This combination allows the determination of impact vibration characteristics to no longer rely solely on energy magnitude, but to comprehensively consider the impact pattern and aperiodicity of the signal. For example, a high-energy signal exhibiting obvious periodicity can be identified as mechanical vibration rather than cavitation impact; conversely, a signal with moderate energy but impact characteristics and aperiodicity can be identified as cavitation impact. It is precisely because of this multi-dimensional and comprehensive judgment that this scheme can effectively filter out interference from mechanical vibration and environmental noise, accurately identify and quantify the impact vibration characteristics caused by cavitation in the low-frequency signal segment. Furthermore, this scheme is closely integrated with the previous step of extracting low-frequency signal segments synchronized with the high-frequency transient events from the low-frequency characteristic signals based on the time point of the high-frequency transient events. The previous step ensures that the analyzed low-frequency signal segments are highly aligned with the high-frequency transient events in time, thereby guaranteeing that the low-frequency signal segments contain potential information related to cavitation. Based on this, this scheme uses impact vibration characteristic analysis and aperiodic analysis, combined with energy distribution, to deeply mine and finely identify the synchronized low-frequency signal band. This combination of time-domain synchronization and multi-dimensional feature analysis improves the accuracy and robustness of cavitation feature extraction, providing a basis for subsequent judgment of the cavitation state of the pure water hydraulic pump valve.

[0084] In practical implementation, the impact vibration characteristics of the low-frequency signal band can be determined as follows: First, by analyzing the impact vibration characteristics of the low-frequency signal band, the kurtosis and crazing factor of the signal band can be calculated. The kurtosis reflects the impulsiveness of the signal, while the crazing factor measures the instantaneous impact intensity of the signal. Simultaneously, aperiodic analysis can be performed to calculate the autocorrelation function of the low-frequency signal band and analyze its decay characteristics at non-zero delays. If the autocorrelation function decays rapidly to near zero, it indicates that the signal has strong aperiodicity; conversely, if there are obvious periodic peaks, it indicates that the signal contains periodic components. Through these analyses, a second analysis result can be obtained, such as an eigenvector containing the kurtosis, crazing factor, and autocorrelation function decay rate. Subsequently, the energy distribution of the low-frequency signal band is obtained, which can be achieved by calculating the root mean square value or the total energy integral of the signal band. Finally, the obtained energy distribution is combined with the second analysis result to determine the impact vibration characteristics of the low-frequency signal band. For example, a rule-based decision logic can be used: if the kurtosis and peak factor of a low-frequency signal band are both higher than a preset threshold, and its autocorrelation function decay rate indicates strong non-periodicity, while its energy distribution also reaches a certain level, then it can be determined that the low-frequency signal band contains impact vibration characteristics caused by cavitation. Conversely, if the signal energy is high but the autocorrelation function shows obvious periodicity, it can be classified as mechanical vibration, thus effectively distinguishing between cavitation impact and mechanical vibration.

[0085] This solution analyzes the impact vibration characteristics and non-periodic properties of low-frequency signal bands, and combines this with energy distribution analysis to effectively distinguish between random, transient impacts caused by cavitation and periodic or quasi-periodic vibrations generated by internal mechanical motion of pumps and valves, as well as external noise. This allows for the identification and quantification of impact vibration features characterizing cavitation in low-frequency signal bands even in environments with complex signal aliasing, thereby avoiding misjudgments or omissions in cavitation status monitoring and improving the accuracy and robustness of cavitation status monitoring in pure water hydraulic pumps and valves.

[0086] Reference Appendix Figure 2 This invention provides a vibration monitoring system for a pure water hydraulic pump valve, comprising:

[0087] The acquisition module 100 is used to acquire high-frequency acoustic emission signals and low-frequency vibration signals; both high-frequency acoustic emission signals and low-frequency vibration signals are acquired synchronously on the pump valve housing of the pure water hydraulic pump.

[0088] The first processing module 200 is used to process the high-frequency acoustic emission signal to obtain the high-frequency characteristic signal;

[0089] The second processing module 300 is used to process the low-frequency vibration signal to obtain the low-frequency characteristic signal;

[0090] The identification and recording module 400 is used to identify high-frequency transient events based on high-frequency characteristic signals, and to record the time point and transient characteristics of the high-frequency transient events;

[0091] The extraction module 500 is used to extract a low-frequency signal segment that is synchronized with the high-frequency transient event from the low-frequency characteristic signal based on the time point of the high-frequency transient event, and to determine the impact vibration characteristics of the low-frequency signal segment.

[0092] The determination module 600 is used to determine the cavitation state of the pure water hydraulic pump valve based on the transient characteristics of high-frequency transient events and the impact vibration characteristics of low-frequency signal bands.

[0093] In this document, relational terms such as first and second are used only to distinguish one entity or operation from another entity or operation, without necessarily requiring or implying any such actual relationship or order between these entities or operations.

[0094] The above description is merely an embodiment of the present invention and is not intended to limit the scope of protection of the present invention. For those skilled in the art, the present invention can have various modifications and variations. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for monitoring the vibration of a pure water hydraulic pump valve, characterized in that, Includes the following steps: High-frequency acoustic emission signals and low-frequency vibration signals are acquired; both high-frequency acoustic emission signals and low-frequency vibration signals are synchronously acquired on the valve housing of the pure water hydraulic pump. The high-frequency acoustic emission signal is processed to obtain the high-frequency characteristic signal; The low-frequency vibration signal is processed to obtain the low-frequency characteristic signal; Based on high-frequency characteristic signals, identify high-frequency transient events and record the time points and transient characteristics of the high-frequency transient events; identifying high-frequency transient events means determining multiple candidate transient events from high-frequency characteristic signals based on changes in signal energy or amplitude, and judging whether the candidate transient events meet the preset cavitation event characteristics through time-frequency analysis and waveform morphology analysis; Based on the time point of the high-frequency transient event, extract the low-frequency signal segment that is synchronized with the high-frequency transient event from the low-frequency characteristic signal, and determine the impact vibration characteristics of the low-frequency signal segment; extracting the low-frequency signal segment that is synchronized with the high-frequency transient event from the low-frequency characteristic signal means obtaining the low-frequency vibration signal segment that is time-aligned with the high-frequency event from the low-frequency characteristic signal based on the time point of the high-frequency transient event. The cavitation state of the pure water hydraulic pump valve is determined based on the transient characteristics of high-frequency transient events and the impact vibration characteristics of low-frequency signal bands.

2. The method for monitoring vibration of a pure water hydraulic pump valve according to claim 1, characterized in that, The transient characteristics of high-frequency transient events include instantaneous amplitude, instantaneous energy, pulse rate, and energy in a specific frequency band.

3. The method for monitoring vibration of a pure water hydraulic pump valve according to claim 1, characterized in that, The steps for identifying high-frequency transient events based on high-frequency characteristic signals include: From high-frequency characteristic signals, multiple candidate transient events are identified based on changes in signal energy or amplitude, and the start time and duration of each candidate transient event are recorded. By performing time-frequency analysis on each candidate transient event, the time-frequency characteristics of each candidate transient event are obtained; By performing morphological analysis on the waveforms of each candidate transient event, the waveform morphological characteristics of each candidate transient event are obtained. Based on time-frequency characteristics and waveform morphology features, high-frequency transient events are identified by judging whether each candidate transient event conforms to the preset cavitation event characteristics.

4. The method for monitoring the vibration of a pure water hydraulic pump valve according to claim 3, characterized in that, Time-frequency characteristics include the energy distribution and duration of candidate transient events within a specific frequency range.

5. The method for monitoring vibration of a pure water hydraulic pump valve according to claim 3, characterized in that, Waveform morphology features include the pulse width, rise steepness, and decay characteristics of candidate transient events.

6. The method for monitoring vibration of a pure water hydraulic pump valve according to claim 1, characterized in that, Based on the time point of the high-frequency transient event, the steps for extracting the low-frequency signal segment synchronized with the high-frequency transient event from the low-frequency characteristic signal include: Based on the transient characteristics of high-frequency transient events, the propagation response characteristics of high-frequency transient events in pure water medium and pump valve structure are obtained by analyzing the energy decay characteristics and duration of high-frequency transient events. The initial time window is determined based on the propagation response characteristics. The initial time window is centered on the time point of the high-frequency transient event, and its length is adjusted according to the duration of the propagation response characteristics. Within the initial time window, the first analysis result is obtained by performing cross-correlation analysis on the envelope signal and low-frequency characteristic signal of the high-frequency transient event; Based on the results of the first analysis, a precisely aligned time window is formed by determining the start and end times of the low-frequency signal band. Low-frequency signal segments are extracted from low-frequency characteristic signals based on precisely aligned time windows.

7. The method for monitoring vibration of a pure water hydraulic pump valve according to claim 6, characterized in that, Within the initial time window, the steps to obtain the first analysis result by performing cross-correlation analysis on the envelope signal and low-frequency characteristic signal of the high-frequency transient event include: Within the initial time window, cross-correlation is performed on the envelope signal of the high-frequency transient event and the low-frequency characteristic signal to obtain the cross-correlation function; Identify multiple cross-correlation peaks from the cross-correlation function, and obtain the time offset and peak intensity of each cross-correlation peak; The morphological characteristics of each cross-correlation peak are obtained by performing morphological analysis on each cross-correlation peak. By combining the transient characteristics of high-frequency transient events, and by comprehensively judging the time offset, peak intensity and morphological characteristics of cross-correlation peaks, cross-correlation peaks that conform to the unique correlation pattern of cavitation events can be identified. Based on the identified cross-correlation peaks, the precise time delay information between the high-frequency transient event and the low-frequency characteristic signal is determined and used as the first analysis result.

8. The method for monitoring the vibration of a pure water hydraulic pump valve according to claim 7, characterized in that, Morphological characteristics include the width, symmetry, and tailing properties of the cross-correlation peaks.

9. The method for monitoring vibration of a pure water hydraulic pump valve according to claim 1, characterized in that, The impact vibration characteristics of the low-frequency signal band are determined according to the following steps: The second analysis result was obtained by performing impact vibration characteristic analysis and non-periodic analysis on the low-frequency signal band; The energy distribution of the low-frequency signal band is obtained, and the impact vibration characteristics of the low-frequency signal band are determined by combining the second analysis results.

10. A vibration monitoring system for a pure water hydraulic pump valve, characterized in that, include: The acquisition module is used to acquire high-frequency acoustic emission signals and low-frequency vibration signals; both high-frequency acoustic emission signals and low-frequency vibration signals are synchronously acquired from the pump valve housing of the pure water hydraulic pump. The first processing module is used to process the high-frequency acoustic emission signal to obtain the high-frequency characteristic signal; The second processing module is used to process the low-frequency vibration signal to obtain the low-frequency characteristic signal; The identification and recording module is used to identify high-frequency transient events based on high-frequency characteristic signals and record the time points and transient characteristics of the high-frequency transient events. Identifying high-frequency transient events means determining multiple candidate transient events from high-frequency characteristic signals based on changes in signal energy or amplitude, and judging whether the candidate transient events meet the preset cavitation event characteristics through time-frequency analysis and waveform morphology analysis. The extraction module is used to extract a low-frequency signal segment that is synchronized with the high-frequency transient event from the low-frequency characteristic signal based on the time point of the high-frequency transient event, and to determine the impact vibration characteristics of the low-frequency signal segment. Extracting a low-frequency signal segment that is synchronized with the high-frequency transient event from the low-frequency characteristic signal refers to obtaining a low-frequency vibration signal segment that is time-aligned with the high-frequency event from the low-frequency characteristic signal based on the time point of the high-frequency transient event. The determination module is used to determine the cavitation state of the pure water hydraulic pump valve based on the transient characteristics of high-frequency transient events and the impact vibration characteristics of low-frequency signal bands.

Images