Seal condition on-line monitoring and leakage identification method and system for rotary machines

Abstract

The present application belongs to the field of seal identification, and particularly relates to a seal state online monitoring and leakage identification method and system for a rotary machine, which comprises the following steps: acquiring a pump outlet working pressure, a micro-environment pressure of an oil seal front cavity and an oil seal seat end surface vibration time domain waveform; calculating a real-time effective pressure difference and combining an allowable shaft seal inlet pressure threshold curve to determine oil deficiency critical wear and record a first trigger mark; performing envelope demodulation on the vibration waveform, determining local material peeling based on a non-integer multiple growth relationship between an impact characteristic value and a rotation frequency, and recording a second trigger mark; after the above marks are generated, continuously monitoring a micro-environment pressure time drift slope, and when the drift slope indicates that the pressure irreversibly approaches an atmospheric pressure direction in an exponential form, determining that a through micro-channel has been formed, the seal interface is at a macroscopic leakage evolution end point, and a forced shutdown maintenance instruction is immediately generated, so that failure early warning is realized before dripping occurs.

Description

Technical Field

[0001] This invention belongs to the field of seal identification, and particularly relates to a method and system for online monitoring of seal status and leakage identification of rotating machinery. Background Technology

[0002] As a core power component of hydraulic lubrication systems, gear pumps are equipped with a skeletonized rotary oil seal at their shaft ends to prevent leakage of internal high-pressure media along the rotating shaft. In long-cycle operation scenarios such as wind power lubrication, construction machinery, and industrial hydraulics, the reliability of this oil seal directly determines the maintenance-free operating time of the system. Existing maintenance strategies rely on periodic inspections for leak detection or passive shutdown and replacement after obvious external leaks occur. However, the failure mechanism of skeletonized oil seals made of nitrile rubber or fluororubber exhibits significant thermo-mechanical coupling evolution characteristics: the oil seal lip and the high-speed rotating alloy steel journal continuously generate heat through friction under a specific pressure difference; when the medium viscosity decreases due to temperature rise, causing the lubricating film to thin, the suction side negative pressure increases due to filter blockage, or the system pressure approaches the allowable pressure difference limit, the lip temperature rises sharply, leading to material aging, hardening, and microcrack propagation. This micro-wear process often lasts for hundreds of hours before macro-leakage occurs. However, existing methods fail to detect this process by utilizing the micro-environmental pressure and pump body vibration time-domain waveform of the chamber between the oil seal and the front bearing. Operational data such as pump body outlet pressure, flow rate, and casing vibration are disconnected from the micro-environmental state such as the concentration of oil and gas escaping from the sealing chamber. This prevents maintenance personnel from intervening in the early stages when the lip enters critical wear due to lack of oil or when local micro-material peeling occurs. Ultimately, sudden oil leakage and shutdown occur due to the formation of a through-channel microchannel, which seriously restricts the high-reliability operation of critical fluid equipment.

[0003] Therefore, the core technical problem to be solved by this application is: how to identify the entire process of leakage evolution of the rotary oil seal lip under the action of thermo-mechanical coupling, from critical wear due to lack of oil, local micro-material peeling, to the formation of through-channel microchannels, so as to achieve proactive early warning when the lip elastomer shows signs of aging, hardening, or micro-crack propagation but before macroscopic dripping, so as to break through the bottleneck of sudden oil leakage shutdown caused by the existing periodic inspection and passive shutdown replacement mode. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention proposes a method and system for online monitoring of sealing conditions and leakage identification in rotating machinery. The method includes: acquiring the pump outlet working pressure, the microenvironment pressure in the oil seal's front chamber, and the time-domain waveform of vibration on the oil seal seat end face; calculating the real-time effective pressure difference and, in conjunction with the allowable shaft seal inlet pressure threshold curve, determining critical wear due to oil exhaust and recording a first trigger mark; performing envelope demodulation on the vibration waveform, and, based on the non-integer multiple growth relationship between the impact characteristic value and the rotational frequency, determining local material spalling and recording a second trigger mark; after the above marks are generated, continuously monitoring the time drift slope of the microenvironment pressure; when the drift slope indicates an irreversible exponential approach of pressure from negative pressure to atmospheric pressure, determining that a permeable microchannel has formed and the sealing interface is at the endpoint of macroscopic leakage evolution, then generating a forced shutdown maintenance command, thereby achieving a failure warning before leakage occurs.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] A method for online monitoring of sealing status and leakage identification of rotating machinery, applied to a fluid transport system including an oil delivery gear pump, wherein a skeleton rotary oil seal is installed between the rotating shaft and the housing of the gear pump, characterized in that it includes:

[0007] The working pressure of the pump outlet pipeline, the microenvironment pressure of the chamber between the oil seal and the front bearing, and the time-domain waveform of the pump body vibration at the end face of the oil seal seat are obtained.

[0008] The real-time effective pressure difference is calculated based on the working pressure and the micro-environment pressure, and a preset allowable shaft seal inlet pressure threshold curve is obtained based on the current operating speed and medium viscosity. When the micro-environment pressure is lower than the lower limit of the allowable shaft seal inlet pressure threshold curve and continues for a first preset time, the oil seal lip lubrication state is determined to have entered the oil shortage critical wear stage, and a first trigger mark is recorded.

[0009] Envelope demodulation is performed on the vibration time-domain waveform to extract the repetitive impact characteristic value of the preset frequency band. When the impact characteristic value exceeds the corresponding vibration intensity baseline and shows an increasing trend that is not an integer multiple of the current frequency, and the real-time effective differential pressure display shows stable operating conditions, it is determined that local micro-material peeling has occurred at the oil seal lip, and a second trigger mark is recorded.

[0010] Within a second preset time period after the first or second trigger mark is generated, the time drift slope of the microenvironment pressure is continuously calculated. If the drift slope indicates that the microenvironment pressure is irreversibly and exponentially approaching atmospheric pressure from a negative pressure state, it is determined that the airtight contact zone of the oil seal lip has formed a through-channel microchannel, and the sealing interface is in a state before failure, evolving from micro-wear to macro-leakage, and a forced shutdown maintenance command is generated.

[0011] Specifically, the shaft seal inlet pressure threshold curve and the vibration intensity baseline are allowed to be stored in a seal failure correlation database containing different speed ranges, working pressure differences and media viscosity combinations. The database also stores a theoretical internal leakage rate drift model per unit time.

[0012] When the first trigger mark is recorded, a first-level warning instruction is also generated to prompt adjustment of the inhalation tubing conditions; when the second trigger mark is recorded, a second-level warning instruction is also generated.

[0013] The preset frequency band is from 2kHz to 10kHz, the vibration intensity baseline is the characteristic frequency band vibration intensity baseline of the preset measuring point of the shell, and the operating condition is stable, specifically, the fluctuation amplitude of the real-time effective pressure difference is within the preset stable threshold range.

[0014] Specifically, the determination that the oil seal lip has entered the critical wear stage due to insufficient oil lubrication is as follows:

[0015] Based on the real-time acquired current operating speed and current medium viscosity, a joint search is performed in a preset seal failure association database, and a dynamic allowable lower limit value of shaft seal inlet pressure matching the current operating conditions is obtained through two-dimensional interpolation calculation; wherein, the seal failure association database stores allowable lower limit values ​​of shaft seal inlet pressure under different speed ranges and different medium viscosity combinations, which are pre-calibrated through bench tests;

[0016] The real-time collected microenvironment pressure value is compared with the lower limit of the dynamically permissible shaft seal inlet pressure.

[0017] When the microenvironment pressure value is lower than the lower limit of the dynamic allowable shaft seal inlet pressure, the first preset duration timer is started.

[0018] If the microenvironment pressure value rises to a level greater than or equal to the lower limit of the dynamic allowable shaft seal inlet pressure within the first preset time period, it is determined that the microenvironment pressure value collected at the corresponding time exceeds the limit due to instantaneous fluctuations in the working condition. The first preset time period is then reset to zero, and the timing is restarted when the microenvironment pressure value falls below the lower limit of the dynamic allowable shaft seal inlet pressure again.

[0019] When the cumulative duration for which the microenvironment pressure value is continuously lower than the lower limit of the dynamic allowable shaft seal inlet pressure reaches the first preset duration, it is determined that the oil seal lip has entered the critical wear stage due to the continuous thinning of the lubricating film, and the first trigger mark is recorded.

[0020] Specifically, the vibration time-domain waveform is subjected to envelope demodulation processing to extract repetitive impact characteristic values ​​of a preset frequency band, including:

[0021] The vibration time-domain waveform acquired by the broadband vibration sensor attached to the end face of the rear cover or oil seal seat is subjected to bandpass filtering with a passband frequency of 2kHz to 10kHz to obtain the filtered vibration signal. The 2kHz to 10kHz passband covers the resonant response frequency band excited by the impact of the peeling of the micromaterial of the oil seal lip on the pump body structure.

[0022] The filtered vibration signal is subjected to Hilbert transform, and the amplitude envelope of the analytical signal is calculated to obtain the impact pulse envelope signal;

[0023] For the impact pulse envelope signal, peak detection is performed within a preset analysis time window. Local peak points with amplitudes exceeding a preset trigger level are identified as impact events, and the timestamps and amplitudes of each impact event are recorded.

[0024] Specifically, the process of envelope demodulating the vibration time-domain waveform and extracting repetitive impact characteristic values ​​of a preset frequency band further includes:

[0025] Calculate the root mean square value of the amplitude of all impact events within the analysis time window, and use it as the current impact characteristic value;

[0026] The current impact characteristic value is compared with the characteristic frequency band vibration intensity baseline stored in the seal failure association database. The characteristic frequency band vibration intensity baseline is the root mean square value of the impact characteristic value measured under the current speed and current working pressure difference combination conditions when the oil seal is in a healthy state.

[0027] When the current impact characteristic value is greater than the characteristic frequency band vibration intensity baseline, the coefficient of variation of the time interval between each adjacent impact event within the analysis time window is calculated. The coefficient of variation is the ratio of the standard deviation of the time interval to the mean of the time interval.

[0028] Specifically, the process of envelope demodulating the vibration time-domain waveform and extracting repetitive impact characteristic values ​​of a preset frequency band further includes:

[0029] When the coefficient of variation is greater than a preset non-periodic threshold, the time interval of the impact event is determined to be unevenly distributed, and the increase of the current impact characteristic value relative to the vibration intensity baseline is not an integer multiple of the current frequency.

[0030] Simultaneously, the real-time effective pressure difference calculated based on the pump outlet working pressure and the micro-environment pressure is obtained. When the fluctuation range of the real-time effective pressure difference is within the preset stable threshold range, the current operating condition is determined to be stable.

[0031] When the current impact characteristic value is greater than the vibration intensity baseline, the coefficient of variation is greater than the non-periodic threshold, and the fluctuation amplitude is within the preset stability threshold range, it is determined that local micromaterial peeling has occurred at the oil seal lip, and a second trigger mark is recorded.

[0032] Specifically, the amplitude envelope of the analytic signal is calculated to obtain the impulse pulse envelope signal, which includes:

[0033] The filtered vibration signal obtained after the bandpass filtering is processed by the frequency domain Hilbert transform algorithm based on the fast Fourier transform to obtain an orthogonal signal that is 90 degrees out of phase with the filtered vibration signal.

[0034] The discrete sampling sequence of the filtered vibration signal is taken as the real part sequence, and the orthogonal signal is taken as the imaginary part sequence. An analytical signal is constructed point by point. The value of the analytical signal at the nth sampling point is the sum of the nth value of the real part sequence and the nth value of the imaginary part sequence multiplied by the imaginary unit j.

[0035] The instantaneous amplitude is calculated point by point for the analytical signal. The instantaneous amplitude at the nth sampling point is equal to the square root of the sum of the square of the nth value of the real part sequence and the square of the nth value of the imaginary part sequence.

[0036] The instantaneous amplitude sequence, which is composed of the instantaneous amplitudes of all sampling points arranged in chronological order, is used as the impulse pulse envelope signal.

[0037] Specifically, determining that the airtight contact zone of the oil seal lip has formed a through-channel includes:

[0038] After the first trigger flag or the second trigger flag is generated, the second preset duration timer is started, and within the second preset duration, the microenvironment pressure value is continuously acquired at a preset sampling period. The microenvironment pressure values ​​arranged in the order of sampling time constitute the microenvironment pressure time series.

[0039] The microenvironment pressure time series is fitted with an exponential function, and the fitting function is in the form of P_seal(t)=a×e(b×t)+c, where t is the time variable with the starting time of the second preset duration as zero, P_seal(t) is the microenvironment pressure fitting value at time t, a, b, and c are fitting coefficients determined by the least squares method, b is the exponential growth coefficient characterizing the microenvironment pressure drift rate, and c is the asymptotic value of the fitting function in the direction of atmospheric pressure.

[0040] Calculate the goodness of fit R² of the exponential function and the exponential growth coefficient b. When the goodness of fit R² of the exponential function is greater than the preset goodness of fit threshold and the exponential growth coefficient b is positive, it is determined that the microenvironment pressure time series shows an irreversible drift trend approaching zero pressure value at an exponential rate.

[0041] Specifically, determining that the airtight contact zone of the oil seal lip has formed a through-channel also includes:

[0042] If, before the end of the second preset time period, the microenvironment pressure time series shows an irreversible drift trend approaching zero pressure value at an exponential rate, it is continuously monitored and confirmed that the oil seal lip airtight contact zone has formed a through-channel microchannel and the sealing interface is in a state of failure before evolving from micro wear to macro leakage, and a forced shutdown maintenance command is generated.

[0043] If, at the end of the second preset duration, the goodness of fit R² of the exponential function is less than or equal to the preset goodness of fit threshold, or the exponential growth coefficient b is negative or zero, then it is determined that the change in the microenvironment pressure still belongs to normal operating condition fluctuations or that the oil seal deterioration has not yet entered the leakage evolution endpoint stage. The second preset duration timer that has been started is cleared, and the online monitoring process continues to be executed.

[0044] Online monitoring and leak detection systems for the sealing condition of rotating machinery include:

[0045] The acquisition module is configured to acquire the working pressure of the pump outlet pipeline, the microenvironmental pressure of the chamber between the oil seal and the front bearing, and the time-domain waveform of the pump body vibration at the end face of the oil seal seat.

[0046] The critical wear determination module is configured to: calculate the real-time effective pressure difference based on the working pressure and the micro-environment pressure, and obtain a preset allowable shaft seal inlet pressure threshold curve based on the current operating speed and medium viscosity; when the micro-environment pressure is lower than the lower limit of the allowable shaft seal inlet pressure threshold curve and continues for a first preset time, determine that the oil seal lip lubrication state has entered the oil shortage critical wear stage, and record the first trigger mark.

[0047] The peeling identification module is configured to: perform envelope demodulation processing on the vibration time-domain waveform, extract the repetitive impact characteristic value of the preset frequency band, and when the impact characteristic value exceeds the corresponding vibration intensity baseline and shows an increasing trend that is not an integer multiple of the current frequency, and the real-time effective differential pressure display shows stable operating conditions, determine that local micro-material peeling has occurred at the oil seal lip, and record the second trigger mark.

[0048] The decision module is configured to: continuously calculate the time drift slope of the microenvironment pressure within a second preset time period after the first trigger mark or the second trigger mark is generated; if the drift slope indicates that the microenvironment pressure is irreversibly exponentially approaching atmospheric pressure from a negative pressure state, then it is determined that the airtight contact zone of the oil seal lip has formed a through-channel microchannel, the sealing interface is in a pre-failure state evolving from micro wear to macro leakage, and a forced shutdown maintenance command is generated.

[0049] Compared with the prior art, the beneficial effects of the present invention are:

[0050] This invention achieves early identification of the entire leakage evolution process of rotary oil seals by simultaneously acquiring the microenvironmental pressure of the oil seal cavity and the time-domain waveform of pump body vibration, and correlating these two with the pump outlet working pressure. Specifically, by comparing the microenvironmental pressure with the allowable shaft seal inlet pressure threshold curve, it can be determined that the lip has entered the critical wear stage due to excessive suction negative pressure, thus achieving early detection of abnormal lubrication status. By performing envelope demodulation on the vibration waveform and extracting impact characteristic values ​​that are not integer multiples of the rotational frequency, it is possible to identify local micro-material spalling at the lip under stable operating conditions and capture the initial signal of material degradation. Based on this, by continuously monitoring the time drift slope of the microenvironmental pressure, when it irreversibly and exponentially approaches atmospheric pressure, it can be determined that a through-channel microchannel has been formed, and a forced shutdown maintenance command can be issued before macroscopic leakage occurs. This method integrates pump operating parameters and the microenvironment of the sealing cavity from data silos into a unified monitoring system, establishing a dynamic coupling relationship between micro-wear, material spalling and leakage. This breaks through the limitations of the existing periodic leak detection and passive shutdown replacement mode, effectively eliminates the bottleneck of sudden oil leakage shutdown, and significantly improves the operational reliability of critical fluid equipment. Attached Figure Description

[0051] Figure 1 This is a flowchart of the online monitoring and leakage identification method for the sealing status of rotating machinery according to Embodiment 1 of the present invention;

[0052] Figure 2 This is a flowchart illustrating the process of determining localized microscopic material peeling at the oil seal lip in Embodiment 1 of the present invention.

[0053] Figure 3 This is a block diagram of the online monitoring and leakage identification system for the sealing status of rotating machinery according to Embodiment 3 of the present invention; Detailed Implementation

[0054] Example 1

[0055] Please see Figure 1The present invention provides an embodiment of a method for online monitoring of sealing status and leakage identification of rotating machinery, applied to a fluid transport system including an oil delivery gear pump, wherein a skeleton rotary oil seal is installed between the rotating shaft and the housing of the gear pump, and the steps include:

[0056] S1. A database of pre-set pump operating characteristic parameters and seal failure correlation, wherein the database includes at least the allowable shaft seal inlet pressure threshold curve under different speed ranges, working pressure differences and medium viscosity combinations, and the characteristic frequency band vibration intensity baseline of the pre-set measuring points of the casing.

[0057] In this embodiment, the specific implementation of the preset pump operating characteristic parameters and seal failure association database mentioned in S1 is as follows: The seal failure association database is constructed in the offline calibration stage before the online monitoring system is put into operation. It is pre-established by those skilled in the art based on the specific specifications of the radial shaft seal rings made of NBR (nitrile butadiene rubber) or FKM (fluororubber) material installed on the SGP series oil transfer gear pump through a combination of bench tests and numerical simulation, and stored in the non-volatile memory of the online monitoring system for real-time retrieval during the online monitoring stage.

[0058] It should be further explained that the preset method for the allowable shaft seal inlet pressure threshold curve in this embodiment is as follows: during the bench test stage, for the specific pump models covered by the SGP series gear pumps, namely 2.5, 063, 080, 100, 112, 125, 150, 180, and 200 cubic centimeters per revolution, the corresponding skeleton rotary oil seal specifications are installed respectively. The speed node values ​​for bench tests are determined based on the speed range given in the SGP sample and the typical speed points in the shaft seal characteristic data table, covering all speed nodes within the ranges of 750 rpm, 1000 rpm, 1500 rpm, 2000 rpm, 2500 rpm, 3000 rpm, and 3600 rpm. At the same time, the maximum speed limits for each displacement specification are matched: SGP2.5~063 maximum speed 3600 rpm, SGP080~180 maximum speed 3000 rpm, and SGP200 maximum speed 2500 rpm. The working differential pressure node values ​​are set based on the upper limit of differential pressure related to the medium viscosity given in the sample allowable differential pressure table. 3 bar is used when the kinematic viscosity of the medium is greater than or equal to 1.4 mm² / s, 12 bar is used when the kinematic viscosity of the medium is greater than or equal to 6 mm² / s, and 25 bar is used when the kinematic viscosity of the medium is greater than or equal to 12 mm² / s. Multiple differential pressure nodes are set in conjunction with the allowable pressure range on the suction side at various speeds in the shaft seal characteristic data table. The medium viscosity node values ​​are set based on the viscosity range of 1.4 mm² / s to 100,000 mm² / s given in the sample and the viscosity-temperature characteristics of three typical hydraulic oils: ISO VG32, VG46, and VG68. Within the working temperature range of 20°C to 80°C, the corresponding kinematic viscosity values ​​are calculated at 5-degree Celsius intervals. For each operating condition combination orthogonally formed by the above-mentioned rotational speed, operating pressure difference, and medium viscosity node values, after maintaining stable rotational speed, pressure difference, and medium temperature, the suction side negative pressure is gradually increased at a gradual rate of approximately 0.05 bar per minute. This causes the micro-environment pressure in the oil seal pre-cavity to gradually decrease from the allowable pressure range given in the sample shaft seal characteristic data table towards the lower limit of negative pressure, until the critical wear characteristic of oil exhaust is triggered. The micro-environment pressure value at the trigger moment is recorded as the allowable lower limit value of the shaft seal inlet pressure under that operating condition combination. After all operating conditions are calibrated, surface fitting is performed on the discrete lower limit values ​​to generate an allowable shaft seal inlet pressure threshold curve with rotational speed and medium viscosity as independent variables, which is stored in the seal failure correlation database.

[0059] It should be further explained that the preset method for the characteristic frequency band vibration intensity baseline of the housing preset measurement point in this embodiment is as follows: Under the condition that the oil seal is in a healthy state, a wideband vibration sensor is attached to the preset measurement point on the housing of the SGP series gear pump, either on the rear cover or the end face of the oil seal seat. The frequency response range of the sensor covers no less than 10 kHz. Under steady-state conditions of various speed ranges and working pressure difference combinations covered by the database, the pump is run and vibration time-domain waveforms are collected, with a collection time of no less than 10 seconds for each condition. The collected vibration time-domain waveforms are bandpass filtered from 2 kHz to 10 kHz. This frequency band is selected to avoid the normal meshing frequency and its low-order harmonics of the SGP series helical gear pump, while effectively covering the resonance response frequency band excited by the impact signal generated when local micro-material peeling occurs at the oil seal lip on the pump body structure. The filtered signal is subjected to Hilbert transform envelope demodulation processing to extract repetitive impact characteristic values. The root mean square value of the impact characteristic values ​​under each steady-state condition is calculated as the vibration intensity baseline value corresponding to that condition. A two-dimensional lookup table is established using rotational speed and pressure difference as indexes for the baseline values ​​of vibration intensity under each working condition, and stored in the seal failure association database. In this embodiment, the preset measurement point on the housing is determined by fixing a broadband vibration sensor with a frequency response range of not less than 10 kHz at a rigid structure near the oil seal mounting stop on the rear cover of the gear pump or the end face of the oil seal seat using a magnetic or threaded mounting method, under the condition that the oil seal is in a healthy state. The pump is operated under steady-state conditions in each speed range and working pressure difference combination covered by the seal failure association database, and vibration time-domain waveforms of not less than 10 seconds are collected. After bandpass filtering of the waveform with a passband frequency of 2 kHz to 10 kHz, the amplitude envelope of the analytical signal is calculated by Hilbert transform and the repetitive impact feature value is extracted. The root mean square value of the impact feature value under each steady-state condition is used as the vibration intensity baseline value of the measurement point corresponding to the current speed and pressure difference combination. Finally, the baseline values ​​of all working condition nodes are used to construct a two-dimensional lookup table indexed by speed and pressure difference, and stored in the seal failure association database for real-time retrieval and retrieval during the online monitoring phase.

[0060] As a specific example, in this embodiment, the rotational speed dimension node of the seal failure association database is set to 750 rpm, 1000 rpm, 1500 rpm, 2000 rpm, 2500 rpm, and 3000 rpm; the working pressure difference dimension node is set to 5 bar, 10 bar, 15 bar, 20 bar, and 25 bar; and the medium viscosity dimension node is set according to the kinematic viscosity values ​​corresponding to the three typical hydraulic oils ISO VG32, VG46, and VG68 in 5-degree Celsius intervals within the working temperature range of 20 degrees Celsius to 80 degrees Celsius, covering the viscosity range of 1.4 mm² / s to 100,000 mm² / s specified in the sample. The allowable suction pressure range given in the sample shaft seal characteristic data table, for the front bearing with radial shaft seal specification, is -0.4 bar to 4.0 bar or -0.4 bar to 3.5 bar at 1500 rpm, -0.4 bar to 2.5 bar or -0.4 bar to 2.0 bar at 2500 rpm, -0.4 bar to 2.0 bar or -0.4 bar to 1.5 bar at 3000 rpm, and -0.4 bar to 1.5 bar at 3600 rpm. These allowable pressure values ​​constitute the initial setting range of suction side operating conditions at various speeds in the calibration test. The node values ​​and coverage ranges of the above parameters can be adjusted by those skilled in the art according to the actual nominal displacement specification, operating condition range, and on-site installation conditions of the target pump type to achieve a reasonable balance between database storage size and online retrieval accuracy; where bar represents the pressure unit.

[0061] S2. Through a pressure transmitter deployed on the pump outlet pipeline, a micro differential pressure sensor deployed near the oil seal mounting stop on the pump body, and a broadband vibration sensor attached to the end face of the rear cover or oil seal seat, the working pressure of the pump outlet pipeline, the micro-environment pressure of the chamber between the oil seal and the front bearing, and the vibration time-domain waveform of the pump body at the end face of the oil seal seat are collected in real time.

[0062] As a specific example, in this embodiment, a certain SGP series oil gear pump with a nominal displacement of 080 is used. An FKM (fluororubber) skeleton rotary oil seal is installed between the rotating shaft and the housing, and the front bearing is a G-type fixed structure with a radial shaft seal. A piezoresistive pressure transmitter with a range of 0 to 40 bar and an output signal of 4 to 20 mA is installed on the straight section of the pump outlet pipeline to collect the working pressure of the pump outlet pipeline in real time. Its sampling frequency is set to 1 kHz, meeting the measurement requirement of the maximum working pressure of 25 bar specified in the sample. A 2 mm diameter pressure tap is located near the oil seal mounting stop on the pump body, specifically on the wall of the sealed chamber between the oil seal and the front bearing. This tap is connected via a stainless steel pressure pipe to a micro-differential pressure sensor with a range of -1 bar to 10 bar and an accuracy class of 0.075. The reference end of this micro-differential pressure sensor is open to the atmosphere, used to acquire the micro-environmental pressure of the chamber between the oil seal and the front bearing in real time, fully covering the suction-side pressure range specified in the sample. Near the circumferential rigid structure on the rear cover end face of the pump body, a piezoelectric broadband vibration sensor with a frequency response range of 0.5 Hz to 15 kHz and a sensitivity of 100 mV / g is magnetically mounted. This sensor is used to acquire the time-domain waveform of the pump body's vibration at the oil seal end face in real time, meeting the vibration signal acquisition requirements within the specified speed range of 200–3000 rpm.

[0063] S3. Calculate the real-time effective pressure difference based on the working pressure and the micro-environment pressure, and obtain the corresponding allowable shaft seal inlet pressure threshold curve based on the current operating speed and medium viscosity. When the micro-environment pressure is lower than the lower limit of the allowable shaft seal inlet pressure threshold curve and continues for a first preset time, determine that the oil seal lip lubrication state has entered the oil shortage critical wear stage, record the first trigger mark and generate a first-level early warning command to prompt adjustment of the suction pipeline conditions.

[0064] It should be further explained that this embodiment calculates the real-time effective pressure difference based on the working pressure and the microenvironment pressure, and obtains the corresponding allowable shaft seal inlet pressure threshold curve based on the current operating speed and medium viscosity, specifically including:

[0065] S301. Real-time acquisition of current operating speed and current medium viscosity; the current operating speed is acquired by a speed sensor deployed at the gear pump drive end, or by directly reading the speed feedback value through the communication interface of the frequency converter connected to the gear pump drive motor. The speed range is adapted to the rated range of 0~3000 rpm specified in the sample for the SGP080 specification. The current medium viscosity is acquired by acquiring the current medium temperature in real time through a temperature transmitter deployed on the pump outlet pipeline, and calling the viscosity-temperature characteristic curve or viscosity-temperature relationship fitting formula of the medium pre-stored in the seal failure association database to convert the current medium temperature into the current medium viscosity. The specific form of the medium viscosity-temperature characteristic curve or viscosity-temperature relationship fitting formula adopts the internationally accepted Walther viscosity-temperature relationship model, the expression of which is:

[0066] ;

[0067] Wherein, ν is the kinematic viscosity of the medium, measured in square millimeters per second; T is the absolute temperature of the medium, measured in Kelvin, T = T1 + 273.15, where T1 is the Celsius temperature of the medium; a1 and a2 are fitting coefficients related to the medium type, determined by those skilled in the art during the offline calibration phase before system commissioning by performing least-squares linear regression on the kinematic viscosity data of the target medium measured at different temperature points. During the online monitoring phase, a temperature transmitter deployed on the pump outlet pipeline collects the current Celsius temperature T1 of the medium in real time, converts the units to obtain the absolute temperature T, and substitutes it into the above fitting formula to calculate the current kinematic viscosity ν of the medium. The specific values ​​of the coefficients a1 and a2 in this viscosity-temperature relationship fitting formula can be determined by those skilled in the art through a limited number of tests based on the type and temperature range of the target medium, and stored in the seal failure association database in the form of fitting coefficient pairs. This application does not specifically limit the values ​​of these coefficients.

[0068] S302, using the current operating rotational speed and the current medium viscosity as joint search conditions, perform a two-dimensional interpolation query in the sealing failure association database, and convert to obtain the lower limit value of allowable shaft seal inlet pressure Plim that accurately matches the current operating condition. The specific operation logic of the two-dimensional interpolation query is: compare the current operating rotational speed ωcur with each node value in the rotational speed dimension of the sealing failure association database one by one, and determine the adjacent low rotational speed node ωlow and adjacent high rotational speed node ωhigh that enclose the current operating rotational speed; compare the current medium viscosity ηcur with each node value in the viscosity dimension of the sealing failure association database one by one, and determine the adjacent low viscosity node ηlow and adjacent high viscosity node ηhigh that enclose the current medium viscosity; according to the four lower limit values of allowable shaft seal inlet pressure P(ωlow, ηlow), P(ωhigh, ηlow), P(ωlow, ηhigh) and P(ωhigh, ηhigh) corresponding to the adjacent rotational speed nodes and adjacent viscosity nodes in the database, first perform linear interpolation on the two lower limit values corresponding to the low viscosity node and the two lower limit values corresponding to the high viscosity node in the rotational speed dimension respectively, then perform linear interpolation on the two obtained interpolation results in the viscosity dimension, to obtain the lower limit value of allowable shaft seal inlet pressure Plim that accurately corresponds to the current operating rotational speed and the current medium viscosity; if the current operating rotational speed or the current medium viscosity is exactly equal to the database node value, the corresponding interpolation dimension degenerates to direct value taking, and the calculation process still holds.

[0069] S303, compare the microenvironment pressure value Pseal collected in real time with the lower limit value of allowable shaft seal inlet pressure Plim. When Pseal < Plim, start the first preset duration timer to count time; during the timing process, if Pseal rebounds to be greater than or equal to Plim, it is determined that the out-of-limit microenvironment pressure value collected at the corresponding time is caused by instantaneous working condition fluctuation, reset the timer to zero and restart timing when Pseal is lower than Plim again; when the cumulative duration of Pseal continuously lower than Plim reaches the first preset duration T2, it is determined that the oil seal lip contact area enters boundary lubrication or dry friction state due to lubricating film thinning, the lubrication state of the oil seal lip enters the oil depletion critical wear stage, and the first trigger flag is recorded.

[0070] The first preset duration T2 is set based on the following: In bench testing, transient disturbances in the suction line are simulated, including typical transient events such as a momentary partial blockage of the filter followed by immediate recovery and brief air intake in the suction line. The maximum duration T_transient_max of the short-term pressure drop in the microenvironment caused by each transient event is recorded. The first preset duration T2 is set to be greater than T_transient_max to ensure that the critical wear judgment is triggered only when the negative pressure on the suction side continuously deteriorates rather than during transient disturbances. This also adapts to the short-term negative pressure tolerance characteristic of a maximum of -1 bar at startup specified in the sample. As a specific example, in this embodiment, the maximum duration of transient disturbance calibrated by bench testing is 3 seconds, and the first preset duration T2 is set to 5 seconds. The specific value of the first preset duration can be determined by those skilled in the art through transient disturbance simulation tests based on the suction line characteristics and typical operating conditions of the target pump type.

[0071] As a specific example, in this embodiment, a certain SGP series oil gear pump with a nominal displacement of 080 is used. It is equipped with a rotary oil seal made of FKM (fluororubber) material, and the drive motor is controlled by a frequency converter with a rated maximum speed of 3000 rpm. At a certain online monitoring moment, the current operating speed ω_cur read through the frequency converter communication interface is 1750 rpm; the temperature transmitter deployed on the pump outlet pipeline collects the current medium temperature in real time as 55 degrees Celsius. By calling the ISOVG46 hydraulic oil viscosity-temperature characteristic curve pre-stored in the seal failure correlation database, the current medium kinematic viscosity η_cur is calculated to be 34 mm² / s, which is within the normal operating viscosity range specified in the sample. In the seal failure association database, the speed dimension nodes are set to 750 rpm, 1000 rpm, 1500 rpm, 2000 rpm, 2500 rpm, and 3000 rpm. The viscosity dimension nodes are set according to the kinematic viscosity values ​​of ISO VG32, VG46, and VG68 hydraulic oils in 5-degree Celsius intervals within the range of 20°C to 80°C. Using ω_cur=1750 rpm and η_cur=34 mm² / s as joint search conditions, adjacent low speed nodes ω_low=1500 rpm and adjacent high speed nodes ω_high=2000 rpm are determined in the speed dimension. In the viscosity dimension, viscosity node values ​​of 32 mm² / s and 36 mm² / s are determined to surround the current viscosity value of 34 mm² / s, i.e., η_low=32 mm² / s and η_high=36 mm² / s. Based on the four node combinations, the corresponding permissible lower limits of shaft seal inlet pressure obtained from the database are P(1500,32)=-0.38bar, P(2000,32)=-0.42bar, P(1500,36)=-0.36bar, and P(2000,36)=-0.40bar. First, linear interpolation is performed on the low viscosity and high viscosity nodes along the rotational speed dimension, resulting in P(1750,32)=-0.38+[-0.42-(-0.38)]×(1750-1500) /

[0072] (2000-1500)=-0.40bar, P(1750,36)=-0.36+[-0.40-(-0.36)]×(1750-1500) / (2000-1500)=-0.38bar; Then, linear interpolation is performed on the above two interpolation results along the viscosity dimension, resulting in P_lim=P(1750,34)=-0.40+[-0.38-(-0.40)]

[0073] ×(34-32) / (36-32)=-0.39bar. This value is the lower limit of the allowable shaft seal inlet pressure that precisely matches the current operating conditions, meeting the -0.4bar minimum suction pressure limit requirement specified in the sample shaft seal characteristic data table. The micro-differential pressure sensor deployed near the oil seal mounting stop on the pump body collects the micro-environment pressure value P_seal in real time, which currently reads -0.45bar. Comparing P_seal=-0.45bar with P_lim=-0.39bar, since -0.45bar is lower than -0.39bar, the first preset duration timer is started. In the subsequent 5 consecutive seconds, the microenvironment pressure values ​​sequentially were -0.44 bar, -0.43 bar, -0.46 bar, -0.47 bar, and -0.45 bar, all consistently below P_lim = -0.39 bar. The timer accumulated to the first preset duration T2 = 5 seconds, indicating that the oil seal lip contact area had entered a boundary lubrication or dry friction state due to excessive negative pressure on the suction side causing lubrication film thinning. The oil seal lip lubrication state had entered the critical wear stage due to insufficient oil, and the first trigger marker was recorded, generating a first-level warning command. It should be noted that the numerical values ​​and specific operating parameters in this embodiment are exemplary descriptions to facilitate understanding of the working principle of this solution. In actual implementation, the allowable lower limit of shaft seal inlet pressure, the specific value of the first preset duration, and the fitting parameters of the viscosity-temperature characteristic curve corresponding to each node in the seal failure association database should all be determined by those skilled in the art through bench testing based on the specifications of the target pump type, the model and material of the skeleton rotary oil seal, the operating condition range, and actual operating conditions. This application does not specifically limit these parameters.

[0074] S4. Perform envelope demodulation processing on the vibration time-domain waveform, extract the repetitive impact characteristic value of the preset frequency band. When the impact characteristic value exceeds the vibration intensity baseline of the characteristic frequency band and shows an increasing trend that is not an integer multiple of the current frequency, and the fluctuation amplitude of the real-time effective pressure difference is within the preset stable threshold range, it is determined that local micro-material peeling has occurred at the oil seal lip, the second trigger mark is recorded, and a secondary warning command is generated; the preset frequency band is 2kHz to 10kHz.

[0075] It should be further explained that this embodiment performs envelope demodulation processing on the vibration time-domain waveform, extracts the repetitive impact characteristic values ​​of the preset frequency band, and determines that local micro-material peeling has occurred at the oil seal lip, specifically including:

[0076] S401. The vibration time-domain waveform acquired by the broadband vibration sensor attached to the end face of the rear cover or oil seal seat is subjected to bandpass filtering with a passband frequency of 2kHz to 10kHz to obtain the filtered vibration signal. The 2kHz to 10kHz passband covers the resonant response frequency band excited by the impact of the peeling off of the micromaterial of the oil seal lip on the pump body structure.

[0077] S402. Perform Hilbert transform on the filtered vibration signal, calculate the amplitude envelope of the analytical signal, and obtain the impact pulse envelope signal;

[0078] S403. For the impact pulse envelope signal, perform peak detection within a preset analysis time window, identify local peak points whose amplitude exceeds the preset trigger level as impact events, and record the timestamp and amplitude of each impact event.

[0079] S404. Calculate the root mean square value of the amplitude of all impact events within the analysis time window, and use it as the current impact characteristic value;

[0080] S405. Compare the current impact characteristic value with the characteristic frequency band vibration intensity baseline stored in the seal failure association database. The characteristic frequency band vibration intensity baseline is the root mean square value of the impact characteristic value measured under the current speed and current working pressure difference combination conditions when the oil seal is in a healthy state.

[0081] S406. When the current impact characteristic value is greater than the characteristic frequency band vibration intensity baseline, calculate the coefficient of variation of the time interval between each adjacent impact event within the analysis time window. The coefficient of variation is the ratio of the standard deviation of the time interval to the mean of the time interval.

[0082] S407. When the coefficient of variation is greater than a preset non-periodic threshold, it is determined that the time interval of the impact event is unevenly distributed, and the increase of the current impact characteristic value relative to the vibration intensity baseline is a non-integer multiple of the current frequency.

[0083] S408. Simultaneously acquire the real-time effective pressure difference calculated based on the pump outlet working pressure and the micro-environment pressure. When the fluctuation range of the real-time effective pressure difference is within the preset stable threshold range, determine that the current operating condition is stable.

[0084] S409. When the current impact characteristic value is greater than the vibration intensity baseline, the coefficient of variation is greater than the non-periodic threshold, and the fluctuation amplitude is within the preset stable threshold range, it is determined that local micromaterial peeling has occurred at the oil seal lip, and a second trigger mark is recorded.

[0085] In this embodiment, S401 is implemented as follows: at the beginning of each vibration analysis cycle during the online monitoring phase, the latest vibration time-domain waveform acquired by the broadband vibration sensor attached to the end face of the rear cover or oil seal seat is read from the data acquisition buffer. The waveform length is equal to the preset analysis time window length. Before filtering, the current operating speed is obtained. Based on the current operating speed and the number of teeth of the drive gear of the SGP series gear pump, the current gear meshing frequency and its integer multiple harmonic frequencies are calculated, where the gear meshing frequency is equal to the current operating speed multiplied by the number of teeth of the drive gear and then divided by 60. A preset 2kHz to 10kHz basic filter passband is compared with the integer multiples of the harmonic frequencies of the current gear meshing frequency. When an integer multiple of the harmonic frequency of the current gear meshing frequency falls within the 2kHz to 10kHz basic filter passband, a stopband notch filter is set with that harmonic frequency as the center frequency and a preset notch bandwidth as the stopband width to correct the basic filter passband, resulting in a corrected filter passband. The corrected filter passband removes the frequency range centered on the harmonic frequency and with a bandwidth equal to the preset notch bandwidth from the basic filter passband, thereby preserving the structural resonance response frequency band excited by the micromaterial spalling impact of the oil seal lip while suppressing the periodic impact components of gear meshing that fall within the passband due to speed changes. If none of the integer multiples of the current gear meshing frequency fall within the basic filter passband, the uncorrected basic filter passband is used as the final filter passband. Bandpass digital filtering is applied to the vibration time-domain waveform using either the corrected filter passband or the uncorrected basic filter passband to obtain the filtered vibration signal.

[0086] The lower limit frequency of the basic filter passband, 2kHz, is set based on the fact that the gear meshing frequency and its second harmonic components of the SGP series gear pump are usually concentrated in the frequency band below 2kHz under typical speed range. Setting a high-pass cutoff frequency of 2kHz can effectively suppress the periodic vibration interference generated by gear meshing. The upper limit frequency of the passband, 10kHz, is set based on the fact that the upper limit of the frequency response of the broadband vibration sensor is not less than 10kHz, and the resonance response excited on the pump body structure by the impact signal generated by the peeling of the micromaterial of the oil seal lip is mainly distributed in the frequency band of 2kHz to 10kHz. This frequency band range can be determined by those skilled in the art through modal hammer test or operational modal analysis.

[0087] The preset notch bandwidth is set based on the frequency drift range of gear meshing frequency harmonics when the speed changes slightly. It is recommended that the preset notch bandwidth be ±5% of the center frequency. This value can fully cover the speed fluctuation range under normal operating conditions of the SGP series gear pump, avoiding filter failure due to harmonic frequencies exceeding the notch stopband range caused by small speed changes. The specific value can be further optimized and adjusted by those skilled in the art through experiments based on the speed control accuracy and frequency resolution of the target pump type. The bandpass digital filter is implemented using a finite impulse response bandpass filter or an infinite impulse response bandpass filter. The filter order and stopband attenuation rate are selected by those skilled in the art based on real-time requirements and anti-aliasing needs.

[0088] In this embodiment, S402 is specifically implemented as follows: the filtered vibration signal output by S401 is processed using the frequency domain Hilbert transform algorithm to obtain an orthogonal signal that is phase-orthogonal to the filtered vibration signal; the processing flow of the frequency domain Hilbert transform algorithm is as follows: a fast Fourier transform is performed on the discrete sampling sequence of the filtered vibration signal to obtain a spectrum sequence composed of complex values ​​corresponding to each frequency point; each complex value corresponding to the positive frequency part in the spectrum sequence is multiplied by two; each complex value corresponding to the negative frequency part in the spectrum sequence is set to zero; the complex value corresponding to the zero frequency point in the spectrum sequence remains unchanged; an inverse fast Fourier transform is performed on the spectrum sequence after the above processing; the imaginary part of each complex value in the inverse transform result sequence is taken; and the imaginary parts are arranged in the order of the sampling points to form the orthogonal signal. The discrete sampling sequence of the filtered vibration signal is taken as the real part sequence, and the orthogonal signal is taken as the imaginary part sequence. An analytic signal sequence is constructed point by point. The value of the nth sampling point in the analytic signal sequence is the value of the nth sampling point in the real part sequence plus the product of the value of the nth sampling point in the imaginary part sequence and the imaginary unit. The instantaneous amplitude is calculated point by point for the analytic signal sequence. The instantaneous amplitude of the nth sampling point is equal to the square root of the sum of the squares of the nth sampling point in the real part sequence and the squares of the nth sampling point in the imaginary part sequence. The instantaneous amplitudes calculated from all sampling points are arranged in chronological order to form an instantaneous amplitude sequence, which is then output as the impact pulse envelope signal. The impact pulse envelope signal separates the amplitude modulation component caused by the impact event from the filtered vibration signal, suppresses the steady-state simple harmonic vibration component that persists in the filtered vibration signal, and makes the oil seal lip micro-peeling impact feature, which was originally submerged in the background of gear meshing vibration, stand out, providing an envelope waveform with enhanced signal-to-noise ratio for subsequent impact event identification and feature extraction.

[0089] In this embodiment, S403 is specifically implemented as follows: the impact pulse envelope signal output by S402 is scanned point by point within a preset analysis time window, with the sampling period as the step size. The preset analysis time window is preset as follows: during the bench calibration stage before system commissioning, a person skilled in the art first calculates the reference duration of the analysis time window based on the rated speed range of the target pump type, the number of teeth of the drive gear, and the vibration signal sampling frequency set by the online monitoring system. This duration should at least cover the absolute time corresponding to the oil seal lip completing three complete rotation cycles with the rotating shaft, so as to ensure that a sufficient number of impact events related to the rotation cycle can be captured within the time window. At the same time, based on the sampling frequency, the reference duration is converted into the corresponding number of sampling points, and it is verified whether the number of sampling points meets the lower limit frequency of the bandpass filter passband. If the spectral resolution requirement and the lower limit of the sample size for statistical estimation of impact characteristic values ​​are not met, the number of sampling points will be expanded according to the larger of the two and the final analysis time window length will be calculated in reverse. After calibration, the analysis time window length will be stored as a fixed parameter in the sealing failure association database. It will share the same time window length with the vibration intensity baseline calibration and subsequent impact characteristic extraction in the online monitoring stage. It can be directly retrieved from the database during online monitoring and applied to the interception and processing of vibration time domain waveforms. This will ensure the consistency of analysis conditions throughout the entire life cycle while ensuring the integrity of impact event capture and statistical stability.

[0090] For the currently scanned i-th sampling point, its amplitude is denoted as A(i). The amplitude sequence of the preceding K consecutive sampling points is denoted as {A(iK), A(i-K+1),...,A(i-1)}, and the amplitude sequence of the following K consecutive sampling points is denoted as {A(i+1), A(i+2),...,A(i+K)}, where K is a preset local neighborhood comparison radius, and the value of K is not less than 2. A(i) is numerically compared with the amplitudes of the preceding and following K sampling points one by one: if A(i) is simultaneously greater than the amplitude of each of the preceding K sampling points and simultaneously greater than the amplitude of each of the following K sampling points, then the i-th sampling point is determined to be a local peak point. After determining that it is a local peak point, A(i) is further compared with the preset trigger level: if A(i) is greater than the preset trigger level, the i-th sampling point is identified as the peak point of an impact event, the sampling time of the sampling point is recorded as the timestamp of the impact event, and the amplitude A(i) of the sampling point is recorded as the amplitude of the impact event; if A(i) is less than or equal to the preset trigger level, the local peak point is determined to be a random fluctuation caused by steady-state noise and is not identified as an impact event. The preset trigger level is set as follows: under the condition that the oil seal is in a healthy state, and under the same combination of speed and differential pressure as in the online monitoring stage, the impact pulse envelope signal is obtained according to the same steps in S401 and S402. The amplitude sequence of the envelope signal is calculated point by point within a preset statistical analysis period. The arithmetic mean μ_env and standard deviation σ_env of the amplitude sequence are calculated. The preset trigger level is set to μ_env + M × σ_env, where M is a preset trigger level multiple. The value of M is set so that the probability of the envelope amplitude caused by steady-state noise in the healthy state exceeding the preset trigger level is lower than the preset false trigger probability threshold.In this embodiment, the preset method for the false trigger probability threshold is as follows: During the bench calibration stage before system commissioning, those skilled in the art first confirm that the oil seal is in a healthy state, that is, under the visual inspection of the pump body disassembly, it is confirmed that there is no visible wear, cracks or material peeling on the lip, and the sealing interface is verified to be leak-free through static airtightness testing; under this confirmation condition, at the steady-state operating points of each speed range and working pressure difference combination covered by the seal failure association database, the pump is run one by one, and the sensor layout, signal acquisition parameters, 2 kHz to 10 kHz bandpass filtering and Hilbert envelope demodulation process are completely consistent with those in the online monitoring stage, to obtain the impact pulse envelope signal with a length of not less than the preset statistical analysis period under each operating condition; for the envelope signal amplitude sequence under each operating condition, after removing the transient segments of start-up and shutdown, the arithmetic mean μ_env and standard deviation σ_env of the amplitude are calculated, and the distribution pattern of the amplitude sequence is verified by the normal distribution hypothesis test. If the test fails, the statistical analysis period is extended until the distribution converges. Based on this, according to the general tolerance level for false alarm rate in the industrial monitoring field, and combined with the highest safety integrity level requirement corresponding to the forced shutdown maintenance command targeted by this system, the upper limit of the probability that the envelope amplitude caused by steady-state noise in the healthy state exceeds the trigger level is set to 0.1%. That is, in every thousand trigger level comparisons, the false trigger caused by normal noise fluctuations will not exceed one. This probability value is the preset false trigger probability threshold. Substituting μ_env, σ_env, and the preset false trigger probability threshold into the complementary cumulative distribution function of the Gaussian distribution, the minimum value of the trigger level multiple M is obtained by inverse solution. Based on this, a preset engineering margin is added as the trigger level multiple parameter finally written into the seal failure association database, which can be directly called for the identification of impact events under various working conditions in the online monitoring stage, thereby achieving a quantitative balance between signal detection sensitivity and false trigger suppression capability.

[0091] The preset local neighborhood comparison radius K is positively correlated with the sampling frequency. When the sampling frequency is 10kHz, K is recommended to be 5~10. The preset trigger level multiple M is recommended to be 3~5, so that the probability of the envelope amplitude caused by steady-state noise exceeding the preset trigger level under healthy conditions is less than 0.1%. In this embodiment, S404 is specifically implemented as follows: the root mean square value of the amplitude sequence of all impact events within the analysis time window identified in S403 is calculated, that is, the sum of the squares of the amplitudes of each impact event is divided by the number of impact events and then the square root is taken. The result is used as the current impact feature value. If no impact event is identified within the current analysis time window, the current impact feature value is zero.

[0092] In this embodiment, S405 is implemented as follows: The current operating speed ω_cur and current working pressure difference ΔP_cur corresponding to the current analysis time window are obtained from step S3. The current working pressure difference ΔP_cur is equal to the real-time effective pressure difference calculated in step S3 based on the working pressure of the pump outlet pipeline and the micro-environment pressure. The pressure difference range is adapted to the maximum working pressure difference limit of 25 bar specified in the sample. Using ω_cur and ΔP_cur as joint retrieval variables, a two-dimensional lookup table of the characteristic frequency band vibration intensity baseline is retrieved from the seal failure association database. The two-dimensional lookup table stores the root mean square values ​​of the impact characteristic values ​​measured under the confirmed healthy state of the oil seal, under the combined operating conditions of each speed node and each pressure difference node, following the same processing flow and parameter settings as S401 to S404. During retrieval, firstly, the adjacent low-speed nodes ω_base_low and adjacent high-speed nodes ω_base_high surrounding ω_cur are determined in the speed dimension. Secondly, the adjacent low-pressure-difference nodes ΔP_base_low and adjacent high-pressure-difference nodes ΔP_base_high surrounding ΔP_cur are determined in the pressure-difference dimension. Based on these four nodes, the corresponding four vibration intensity baseline values ​​V(ω_base_low,ΔP_base_low), V(ω_base_high,ΔP_base_low), V(ω_base_low,ΔP_base_high), and V(ω_base_high,ΔP_base_high) are obtained by indexing in the lookup table. These four baseline values ​​are first linearly interpolated in the speed dimension and then linearly interpolated in the pressure-difference dimension to obtain the characteristic frequency band vibration intensity baseline V_baseline that precisely matches the current operating speed ω_cur and the current working pressure difference ΔP_cur. If ω_cur or ΔP_cur is equal to the database node value, the interpolation in the corresponding dimension degenerates into direct value taking. The vibration intensity baseline V_baseline represents the expected amplitude of the impact characteristic value extracted through envelope demodulation under the inherent vibration background level of a healthy oil seal and under the current operating conditions. Subsequently, the current impact characteristic value V_current calculated by S404 is compared with V_baseline. When V_current is greater than V_baseline, it is determined that the current impact characteristic value exceeds the normal fluctuation range under healthy conditions, and the process proceeds to the subsequent non-periodic characteristic determination step.

[0093] In this embodiment, S406 is implemented as follows: When S405 determines that V_current is greater than V_baseline, the timestamp values ​​are extracted from the timestamp sequence of all impact events recorded in S403 within the current analysis time window. The timestamps of each impact event are arranged in the order of sampling time and denoted as T(1), T(2), ..., T(N), where N is the total number of impact events identified within the current analysis time window. When N is less than 2, that is, the number of impact events identified within the current analysis time window is less than two, the time interval between adjacent impact events cannot be calculated. At this time, the subsequent coefficient of variation calculation is not performed, and the determination process of the current analysis cycle is directly terminated. When N is greater than or equal to 2, the time interval between every two adjacent impact events is calculated sequentially. The i-th time interval Δt(i) = T(i+1) - T(i), where i = 1, 2, ..., N-1, and the time interval sequence is composed of all N-1 time intervals. The arithmetic mean Δt_mean = (1 / (N-1)) × Σ_{i=1}^{N-1}Δt(i) is calculated for the time interval sequence, and its standard deviation is calculated. Then, the coefficient of variation CV = Δt_std / Δt_mean. The coefficient of variation CV is a dimensionless value that characterizes the relative dispersion of the time intervals of impact events: when the value of CV approaches zero, it indicates that the time intervals between adjacent impact events tend to be equal, and the occurrence of impact events exhibits a periodic characteristic in time. This periodic characteristic is consistent with the impact mode of deterministic rotational excitation such as gear meshing; when the value of CV increases, it indicates that the difference in the time intervals between adjacent impact events increases, and the occurrence of impact events exhibits a non-uniform distribution in time. This non-periodic characteristic is consistent with the irregular impact mode generated by the random micro-scraping of oil seal lip material during journal rotation.

[0094] In this embodiment, S407 is specifically implemented as follows: the coefficient of variation CV calculated in S406 is numerically compared with the preset non-periodic threshold CV_threshold. The calibration method for the preset non-periodic threshold CV_threshold is as follows: during the bench test phase, firstly, under the condition that the oil seal is confirmed to be in a healthy state, under the same combination of rotational speed and differential pressure as in the online monitoring phase, vibration data for multiple analysis time windows are collected according to the same processing flow and parameter settings as in S401 to S406. The coefficient of variation value is calculated for each analysis time window to obtain a healthy state coefficient of variation sample set, and the statistical upper limit value CV_health_max of the sample set is calculated. Subsequently, the local micro-material peeling state is simulated at the oil seal lip by artificially prefabricating micro-peeling defects, under the same working conditions. Vibration data from multiple analysis time windows were collected using the same processing procedure, and the coefficient of variation was calculated to obtain a sample set of coefficients of variation for the micro-exfoliation state. The statistical lower limit value CV_defect_min of this sample set was calculated. A value between CV_health_max and CV_defect_min was taken as CV_threshold, such that CV_threshold is greater than CV_health_max and less than CV_defect_min, to ensure that the coefficient of variation in the healthy state is lower than CV_threshold and the coefficient of variation in the micro-exfoliation state is higher than CV_threshold. When CV is greater than CV_threshold, it is determined that the time intervals of impact events within the current analysis time window are unevenly distributed, meaning that the occurrence of impact events does not have a periodic characteristic synchronized with the rotational frequency. The increase in the current impact characteristic value V_current relative to the vibration intensity baseline V_baseline originates from the random impact excitation caused by local micro-scraping of the oil seal lip material, rather than the enhancement of the periodic impact generated by gear meshing that is strictly synchronized with the rotational frequency. When CV is less than or equal to CV_threshold, it is determined that the time intervals of impact events within the current analysis time window are evenly distributed, and the occurrence of impact events has a periodic characteristic. The increase in the current impact characteristic value originates from the enhancement of gear meshing impact or other deterministic periodic excitation synchronized with the rotational frequency. The specific value of the preset non-periodic threshold CV_threshold can be determined by those skilled in the art based on the target pump type and specifications, the model and material of the skeleton rotary oil seal, and the statistical distribution results of the coefficient of variation of the healthy state and micro-scraping state in the bench comparison test. This application does not impose specific limitations on this.

[0095] In this embodiment, S408 is specifically implemented as follows: While S406 and S407 are executed, the time series of the real-time effective pressure difference ΔP_cur, calculated based on the working pressure P_out of the pump outlet pipeline and the microenvironmental pressure P_seal between the oil seal and the front bearing chamber, is obtained from step S3. ΔP_cur(t) = P_out(t) - P_seal(t). All sampled values ​​of the real-time effective pressure difference time series within the current analysis time window are extracted. The maximum value ΔP_max and minimum value ΔP_min are obtained by traversing the time series. The difference between the maximum and minimum values, ΔP_fluc = ΔP_max - ΔP_min, is calculated and used as the fluctuation amplitude of the real-time effective pressure difference. The fluctuation amplitude ΔP_fluc is then numerically compared with a preset stability threshold ΔP_stable_threshold. When ΔP_fluc is less than or equal to ΔP_stable_threshold, the current operating condition is determined to be stable, meaning that the pump load and suction side conditions have not changed significantly, and the change in vibration characteristics can be attributed to the change in the state of the oil seal itself rather than external operating condition disturbances. When ΔP_fluc is greater than ΔP_stable_threshold, the current operating condition is determined to be unstable. At this time, even if S407 determines that the coefficient of variation CV is greater than the non-periodic threshold CV_threshold, meaning that the impact characteristic value shows a non-periodic increase, the second trigger flag will not be triggered to avoid misjudgment due to transient abnormalities in vibration characteristics caused by drastic changes in operating conditions. The determination results in the current analysis cycle will be set to invalid, and the determination process of S401 to S408 will be re-executed in the next analysis cycle.

[0096] The calibration method for the preset stability threshold ΔP_stable_threshold is as follows: During the bench test phase, under the condition that the oil seal is in a healthy state, multiple steady-state operating points are sequentially set within the expected operating speed range and expected working differential pressure range of the target pump type. For each steady-state operating point, the normal operating condition fluctuation range is simulated, including differential pressure changes caused by small load adjustments and micro-movements of the suction-side valve. The fluctuation amplitude of the real-time effective differential pressure within multiple analysis time windows under each steady-state operating condition is recorded. All steady-state operating condition fluctuation amplitude values ​​are statistically analyzed, and the upper limit of the statistical analysis is taken as the preset stability threshold ΔP_stable_threshold. This ensures that, under the condition that the oil seal is healthy and the system is operating normally, the probability of the real-time effective differential pressure fluctuation exceeding this threshold is lower than the risk level of misjudgment, while also adapting to the maximum 25 bar working differential pressure range specified in the sample. The specific value of this preset stability threshold can be determined by those skilled in the art based on the operating characteristics of the target pump type, the allowable differential pressure fluctuation range of the process, and the statistical analysis results of the field operating data. This application does not impose specific limitations on this.

[0097] In this embodiment, S409 is specifically implemented as follows: within the same analysis cycle, when the following three conditions are simultaneously met: the comparison result of S405 is V_current greater than V_baseline, the comparison result of S407 is CV greater than CV_threshold, and the comparison result of S408 is ΔP_fluc less than or equal to ΔP_stable_threshold, the generation logic of the second trigger flag is triggered. The simultaneous satisfaction of the above three conditions points from different dimensions to the occurrence of pitting or peeling of local micromaterials at the oil seal lip: Condition 1, V_current greater than V_baseline, indicates that the impact energy in a specific frequency band has exceeded the normal fluctuation range under healthy conditions, and there is an abnormal impact excitation source; Condition 2, CV greater than CV_threshold, indicates that the abnormal impact excitation exhibits non-periodic characteristics in its time distribution, which is different from the periodic impact mode generated by deterministic rotational excitations such as gear meshing; Condition 3, ΔP_fluc less than or equal to ΔP_stable_threshold, excludes the alternative explanation of transient abnormal vibration characteristics caused by drastic fluctuations in operating conditions. The combined determination of these three conditions provides a unique indication of the specific failure mode of localized microscopic material peeling off the oil seal lip. When all three conditions are met simultaneously, the system records the sampling time corresponding to the current analysis time window as the generation time of the second trigger marker, writes the status flag of the second trigger marker into the system status register, and generates a level-two warning command. This command alerts maintenance personnel that localized material peeling has occurred on the oil seal lip, and that the remaining service life of the oil seal and maintenance plan should be comprehensively assessed in conjunction with the trigger status of the level-one warning command and the results of lubricating oil sampling analysis. The lubricating oil filtration accuracy must meet the recommended requirement of no more than 60µm. If any of the above three conditions is not met, the second trigger marker is not triggered, and the system continues to execute the monitoring process from S401 to S409 in the next analysis cycle.

[0098] This embodiment achieves online identification of local microscopic material spalling at the oil seal lip through the above steps. Its technical significance lies in: separating weak impact signals from composite vibrations through Hilbert envelope demodulation; suppressing deterministic frequency interference such as gear meshing through specific frequency bandpass filtering combined with adaptive notch filtering; distinguishing between random micro-spalling impacts and periodic meshing impacts through the calculation of the impact interval variation coefficient; eliminating misjudgments of operating conditions through real-time effective differential pressure fluctuation amplitude judgment; and finally establishing a multi-condition joint judgment mechanism with three independent judgment dimensions: vibration impact characteristic amplitude V_current, impact interval distribution characteristics CV, and operating stability index ΔP_fluc. Each of the three dimensions corresponds to different physical excitation sources and interference source identification functions, and they are independently verified, thereby achieving reliable identification and graded early warning when the lip material spalling is still in the microscopic stage.

[0099] As a specific example, in this embodiment, a certain SGP080 specification oil gear pump has a nominal displacement of 80.50 cubic centimeters per revolution, a 10-tooth drive gear, and a fluororubber-coated rotary oil seal between the rotating shaft and the housing. The front bearing is a G-type fixed structure with a radial shaft seal. At a certain online monitoring moment, the drive motor is controlled by a frequency converter, and the pump operates under steady-state conditions of 1500 revolutions per minute and a working pressure difference of 15 bar, conveying mineral oil with a viscosity of 34 square millimeters per second. According to the SGP sample drive power calculation method, the pump drive power P... Pu =(ω / 1450)×P Tab ×f ν Where the rotational speed ω = 1500 rpm, referring to the sample drive power table, the drive power P corresponding to 1450 rpm and working pressure of 15 bar is found to be... Tab =3.6kW, viscosity coefficient fν=1.0 corresponding to a viscosity of 34 mm² / s, substituting into the formula, we get P Pu =(1500 / 1450)×3.6×1.0≈3.7kW, indicating that the pump is in normal rated operating condition. The corresponding characteristic frequency band vibration intensity baseline V_baseline is pre-calibrated to 0.10g. This baseline value is the root mean square value of the impact characteristic value measured under the condition that the oil seal is in good condition, at the same speed and pressure difference, and according to the same processing procedures and parameter settings as S401 to S404.

[0100] At a certain online monitoring moment, the root mean square value of the current impact characteristic V_current calculated by S404 is 0.18g. S405 determines that 0.18g is greater than 0.10g and proceeds to S406. S406 extracts the total number of impact events N=12 within the current analysis time window from the timestamp sequence recorded in S403. The time interval sequence of each impact event is calculated to have an arithmetic mean Δt_mean = 4.2 milliseconds, a standard deviation Δt_std = 3.8 milliseconds, and a coefficient of variation CV = 3.8 / 4.2 = 0.90. In S407, the preset non-periodic threshold CV_threshold is calibrated to 0.35. It is determined that CV = 0.90 is greater than 0.35, indicating that the impact intervals are unevenly distributed. The increase of the current impact characteristic value V_current relative to the vibration intensity baseline V_baseline originates from the random impact excitation caused by the local microscopic peeling of the oil seal lip material, rather than the enhancement of the periodic impact that is strictly synchronized with the rotation frequency generated by gear meshing.

[0101] S408 acquires the time series of the real-time effective differential pressure ΔP_cur within the current analysis time window, iterates to obtain the maximum value ΔP_max = 15.2 bar and the minimum value ΔP_min = 15.0 bar, and calculates the fluctuation amplitude ΔP_fluc = 15.2 - 15.0 = 0.2 bar. ΔP_fluc = 0.2 bar is compared with the preset stability threshold ΔP_stable_threshold = 0.5 bar. Since 0.2 bar is less than 0.5 bar, the current operating condition is considered stable, and the pump load and suction side conditions have not changed significantly. The preset stability threshold ΔP_stable_threshold = 0.5 bar is calibrated based on the following: In bench testing, under steady-state conditions of healthy oil seals, a speed of 1500 rpm, and a differential pressure of 15 bar, the SGP080 pump simulates small load adjustments and minor movements of the suction side valves caused by normal operation, and the statistical upper limit of the differential pressure fluctuation amplitude is recorded within multiple analysis time windows.

[0102] S409 performs a joint verification of the above three judgment conditions: V_current=0.18g is greater than V_baseline=0.10g, condition one is satisfied; CV=0.90 is greater than CV_threshold=0.35, condition two is satisfied; ΔP_fluc=0.2bar is less than ΔP_stable_threshold=0.5bar, condition three is satisfied. If all three conditions are met simultaneously, the system determines that localized pitting or flaking of micromaterials has occurred on the oil seal lip. At the end of this analysis cycle, a second trigger marker is recorded and a level two early warning command is generated, prompting maintenance personnel that localized material flaking has occurred on the oil seal lip. The remaining service life of the oil seal and the maintenance plan need to be comprehensively evaluated in conjunction with the trigger status of the level one early warning command and the results of lubricating oil sampling analysis.

[0103] It should be noted that the specific values ​​of parameters such as the preset trigger level, characteristic frequency band vibration intensity baseline V_baseline, preset non-periodic threshold CV_threshold, and preset stability threshold ΔP_stable_threshold in this example should be determined by those skilled in the art through bench testing and on-site calibration based on the specifications of the target pump type, the model and material of the rotary oil seal, the operating conditions, and the on-site noise level. This application does not impose specific limitations on these parameters.

[0104] S5. Within a second preset time period after the first trigger mark or the second trigger mark is generated, the time drift slope of the microenvironment pressure is continuously calculated. If the drift slope indicates that the microenvironment pressure is irreversibly and exponentially approaching atmospheric pressure from a negative pressure state, it is determined that the airtight contact strip of the oil seal lip has formed a through-channel microchannel, and the sealing interface is in a state before failure, evolving from micro-wear to macro-leakage. A forced shutdown maintenance command is then generated.

[0105] It should be further explained that S5 in this embodiment specifically includes:

[0106] S501. After the first trigger mark or the second trigger mark is generated, start the second preset duration timing, and within the second preset duration, continuously acquire the microenvironment pressure value at a preset sampling period, and the microenvironment pressure value arranged in the order of sampling time constitutes the microenvironment pressure time sequence.

[0107] S502. Perform exponential function fitting on the microenvironment pressure time series, with the fitting function form being P_seal(t)=a×e (b×t) +c, where t is the time variable with the start time of the second preset duration as zero, P_seal(t) is the microenvironment pressure fitting value at time t, a, b, and c are fitting coefficients determined by the least squares method, b is the exponential growth coefficient characterizing the microenvironment pressure drift rate, and c is the asymptotic value of the fitting function in the direction of atmospheric pressure.

[0108] S503. Calculate the goodness of fit R² of the exponential function and the exponential growth coefficient b. When the goodness of fit R² of the exponential function is greater than the preset goodness of fit threshold and the exponential growth coefficient b is positive, it is determined that the microenvironment pressure time series shows an irreversible drift trend that approaches the direction of atmospheric pressure at an exponential rate.

[0109] S504. If, before the end of the second preset time period, the microenvironment pressure time series shows an irreversible drift trend approaching the zero pressure value at an exponential rate, it has been continuously monitored and confirmed, then it is determined that the airtight contact strip of the oil seal lip has formed a through-channel microchannel, and the sealing interface is in a state of failure before evolving from micro wear to macro leakage, and a forced shutdown maintenance command is generated.

[0110] S505. If, at the end of the second preset time period, the goodness of fit R² of the exponential function is less than or equal to the preset goodness of fit threshold, or the exponential growth coefficient b is negative or zero, then it is determined that the change in the microenvironment pressure still belongs to normal operating condition fluctuations or the oil seal deterioration has not yet entered the leakage evolution endpoint stage. The second preset time period that has been started is cleared, and the online monitoring process from S1 to S4 continues to be executed.

[0111] In this embodiment, S501 is implemented as follows: During online monitoring, when the system status register detects that the flag bit of the first trigger flag or the second trigger flag has been written, the writing time is immediately used as the starting point of the second preset duration timer, and the construction of the microenvironment pressure time series begins. The method of obtaining the microenvironment pressure value is consistent with the microenvironment pressure acquisition method in step S3, that is, it is read in real time from the micro differential pressure sensor deployed near the oil seal mounting stop of the pump body, and the sampling period is the same as the sampling period in step S3. After each sampling of the microenvironment pressure value is completed, the sampled value and its corresponding sampling time are appended to the end of the microenvironment pressure time series. The second preset duration is set based on the following: In bench testing, under the condition of artificially prefabricated through-hole microchannel defects at the oil seal lip, the time elapsed from the formation of the microchannel to the recovery of the microenvironment pressure from a negative pressure state to near atmospheric pressure is measured multiple times. The statistical upper limit of each measurement time is taken, and a preset margin is added to obtain the second preset duration. This ensures that the time window is sufficient to fully capture the complete process of the irreversible exponential approach of the microenvironment pressure from negative pressure to atmospheric pressure. As a specific example, in this embodiment, the maximum time for the microenvironment pressure to recover to near atmospheric pressure after the formation of the through-hole microchannel, as calibrated by bench testing, is 8 minutes. After adding a 2-minute margin, the second preset duration is set to 10 minutes. The specific value of the above-mentioned second preset duration can be determined by those skilled in the art through bench testing calibration based on the sealing cavity volume of the target pump type, the specifications of the rotary oil seal, and the operating conditions. This application does not impose specific limitations on this.

[0112] In this embodiment, S502 is specifically implemented as follows: during the second preset duration of timing, whenever a new sampling point is added to the microenvironmental pressure time series, all currently accumulated microenvironmental pressure sampling values ​​are used as the fitting data source, and an exponential function fitting operation is performed once. The fitting function adopts a three-parameter exponential model P_seal(t)=a×e (b ×t)+c, the fitting algorithm adopts the Levenberg-Marquardt nonlinear least squares method. The specific execution process of this algorithm is as follows: First, initial guess values ​​are assigned to the fitting coefficients a, b, and c. The initial guess value of a is the difference between the first and last sample values ​​of the microenvironment pressure time series. The initial guess value of b is a small positive number. The initial guess value of c is zero or slightly greater than the value of the last sample value. The basis for these values ​​is that, in the oil seal lip penetration microchannel formation stage targeted in this embodiment, the physical evolution law of microenvironment pressure is an irreversible recovery process from negative pressure to atmospheric pressure (i.e., zero gauge pressure). The pressure-time curve of this process shows an exponential growth pattern. b, as the exponential growth coefficient characterizing the microenvironment pressure drift rate, must be positive under the physical mechanism of continuous expansion of the leakage channel; a small positive number (such as 1×10) is taken. -4 Up to 1×10 -3 Using the initial guess (on the order of magnitude of the pressure) allows the Levenberg-Marquardt nonlinear least squares iterative algorithm to start its search from a point close to zero but in the correct direction. This ensures that the convergence direction of the iteration is consistent with the actual physical trend and avoids the possibility of getting trapped in a local optimum due to severe residual oscillations in the early stages of iteration caused by an excessively large initial value. c, as the asymptotic value of the fitting function in the direction of atmospheric pressure, corresponds to the final state of the gauge pressure when the pressure in the sealed cavity reaches equilibrium with the atmospheric pressure after the through-channel microchannel has been fully expanded. When the pressure within the fitting data window has not yet fully recovered to zero, taking zero as the initial guess aligns with the theoretical expectation that the gauge pressure final state is zero. Taking a value slightly larger than the last sampled value is based on the reasonable extrapolation assumption that "the currently observed highest pressure value is slightly lower than the true asymptotic final value." Both methods limit the starting point of the iterative search to a reasonable neighborhood of the final state, thereby accelerating convergence and improving the goodness of fit.

[0113] Then, the optimal coefficient combination that minimizes the sum of squared residuals is obtained through iterative solutions. After each fitting operation, the values ​​of coefficients a, b, and c obtained in that fitting are output. Among them, coefficient b is the exponential growth coefficient. When b is positive, it indicates that the microenvironment pressure drifts exponentially in the direction of increasing value over time, corresponding to the process of the microenvironment pressure approaching the direction of atmospheric pressure from a negative pressure state. For example, the initial guess value of b is 1×10. -4 ~1×10 -3 The small positive number is the exponential drift rate adapted to the microenvironment pressure. The coefficient c is the asymptotic value of the fitting function in the direction of atmospheric pressure. Theoretically, as t approaches infinity, P_seal(t) approaches c, and the value of c should be close to zero or slightly lower than atmospheric pressure.

[0114] In this embodiment, S503 is implemented as follows: After each fitting operation, the goodness of fit R² is calculated. The formula for calculating the goodness of fit R² is R² = 1 - SS_res / SS_tot, where SS_res is the sum of squares of the fitting residuals, i.e., the sum of squares of the differences between the actual measured values ​​of each sampling point in the microenvironmental pressure time series and the calculated values ​​of the fitting function at that sampling time; and SS_tot is the sum of squares of the total deviations, i.e., the sum of squares of the differences between the actual measured values ​​of each sampling point in the microenvironmental pressure time series and the average of the actual measured values ​​of all sampling points. The value of R² ranges from 0 to 1. The closer R² is to 1, the better the exponential function fits the microenvironmental pressure time series. The goodness of fit R² calculated in this operation is compared with a preset goodness of fit threshold R²_threshold, and the exponential growth coefficient b obtained in this operation is compared with zero. When both conditions are met—R² greater than R²_threshold and b being positive—it is determined that the current microenvironmental pressure time series exhibits an irreversible drift trend towards zero pressure at an exponential rate. The preset goodness-of-fit threshold R²_threshold is set based on the following: During bench testing, microenvironmental pressure time series are collected and fitted with an exponential function at the oil seal's critical wear stage (oil exhaustion stage), local micromaterial peeling stage, and through-channel formation stage. The lower limit of the goodness-of-fit for the through-channel formation stage is statistically analyzed, and the preset goodness-of-fit threshold is set above this lower limit, so that a judgment is triggered only when the pressure drift exhibits a significant exponential regularity. As a specific example, in this embodiment, the preset goodness-of-fit threshold R²_threshold is set to 0.85. The specific value of this preset goodness-of-fit threshold can be determined by those skilled in the art based on the statistical distribution characteristics of the bench test data, and this application does not impose specific limitations on it.

[0115] In this embodiment, S504 is implemented as follows: During the timing of the second preset duration, if the condition that the microenvironment pressure time series exhibits an irreversible drift trend approaching zero pressure at an exponential rate is met after the judgment of S503 within a certain sampling period, then S501 to S503 are executed again in the next sampling period to perform exponential function fitting and judgment on the microenvironment pressure time series after the newly added sampling points; if the judgment results of multiple consecutive sampling periods are that the irreversible drift trend condition is met, then it is determined that the irreversible exponential approach trend has been continuously monitored and confirmed. At this time, regardless of whether the second preset duration has been completed, the generation logic of the forced shutdown maintenance command is immediately triggered, determining that the airtight contact strip of the oil seal lip has formed a through-channel microchannel, and the sealing interface is in a state before failure, evolving from micro-wear to macro-leakage. After generating the forced shutdown maintenance command, the system sends the command signal to the control circuit of the pump drive motor through the hard-wired interface or communication bus to perform the shutdown operation, and at the same time writes the diagnostic conclusion that the through-channel microchannel has been formed into the system status register. The specific number of consecutive sampling periods is set based on the following: to avoid misjudgment caused by the random fluctuations of individual sampling points in the microenvironment pressure time series leading to the instantaneous fulfillment of the fitting conditions, the number of consecutive confirmations should not be less than the preset minimum number of consecutive confirmations. As a specific example, in this embodiment, the minimum number of consecutive confirmations is set to 3 times, that is, when the irreversible drift trend condition is determined to be met within 3 consecutive sampling periods, the trend is confirmed as the true trend. The specific value of the minimum number of consecutive confirmations can be determined by those skilled in the art through experiments based on the noise level of the microenvironment pressure signal and the length of the sampling period, and this application does not impose a specific limitation on it.

[0116] In this embodiment, S505 is implemented as follows: When the second preset duration is completed, if the determination result of S503 is always that the goodness of fit R² is less than or equal to the preset goodness of fit threshold R²_threshold, or the exponential growth coefficient b is always negative or zero, that is, the irreversible exponential approximation trend has never been satisfied, then it is determined that within the second preset duration after the generation of the first or second trigger mark, the microenvironment pressure has not undergone an irreversible exponential drift towards atmospheric pressure, the current degree of deterioration of the oil seal lip has not yet entered the leakage evolution endpoint stage of the formation of a through-channel, and the change in microenvironment pressure still belongs to the normal pressure fluctuation range of the critical wear of oil-deficient areas or the local micromaterial peeling stage. At this time, the system clears the started second preset duration timer, resets the timer to zero, and continues to execute the online monitoring process of S1 to S4, waiting for the first or second trigger mark to be triggered again before restarting the determination process of S5.

[0117] This embodiment achieves online identification of the formation of through-channel microchannels at the oil seal lip through the above steps. Its technical significance lies in the fact that when the oil seal deteriorates to the final stage of leakage evolution, the formation of through-channel microchannels establishes a direct connection between the oil seal's front cavity and the atmosphere, causing the microenvironment pressure, originally maintained in a negative pressure state, to recover exponentially towards atmospheric pressure. By fitting an exponential function and calculating the goodness of fit and exponential growth coefficient, this irreversible exponential approximation characteristic can be identified from the pressure time series, distinguishing it from pressure change patterns during normal operating condition fluctuations or other deterioration stages. This allows for timely issuance of forced shutdown maintenance commands before macroscopic leakage occurs, preventing equipment damage and environmental pollution caused by sudden leakage.

[0118] This invention achieves progressive online identification and graded early warning of three key deterioration stages in the entire process of rotary oil seal leakage evolution: critical wear due to lack of oil, local micromaterial peeling, and formation of permeable microchannels. By simultaneously acquiring the microenvironmental pressure of the oil seal front cavity and the time-domain waveform of pump body vibration, and correlating the two with the working pressure of the pump outlet, this invention realizes progressive online identification and graded early warning of three key deterioration stages: critical wear due to lack of oil, local micromaterial peeling, and formation of permeable microchannels. Its beneficial effects are as follows: First, by comparing the microenvironment pressure with the lower limit of the allowable shaft seal inlet pressure threshold curve dynamically obtained from the seal failure correlation database based on the current operating speed and medium viscosity, and combining this with a continuous judgment mechanism for a first preset duration, it can accurately identify the critical wear state of oil shortage at the early stage when the lubricating film thins due to excessive negative pressure on the suction side of the oil seal lip, generating a first-level early warning command, thereby guiding maintenance personnel to adjust the suction pipeline conditions in a timely manner and avoid further deterioration of the lip lubrication state; Second, by performing specific frequency band bandpass filtering and Hilbert envelope demodulation on the vibration time-domain waveform, extracting impact characteristic values ​​and calculating the coefficient of variation of the impact interval, and using the joint judgment of three conditions—impact characteristic value amplitude exceeding the vibration intensity baseline, coefficient of variation greater than the non-periodic threshold, and real-time effective pressure difference fluctuation amplitude within the stable threshold range—it can accurately identify the critical wear state of oil shortage at the early stage when the lubricating film thins due to excessive negative pressure on the suction side of the oil seal lip, generating a first-level early warning command, thereby guiding maintenance personnel to adjust the suction pipeline conditions in a timely manner and avoid further deterioration of the lip lubrication state; Under the background interference of strong gear meshing vibration, it can reliably distinguish between the non-periodic impact caused by random micro-scraping of oil seal lip material and the periodic impact caused by gear meshing, thus achieving automatic identification at the stage of local micro-material spalling and generating a secondary early warning command, providing a basis for the advance formulation of maintenance plans. Finally, after the generation of the first or second trigger mark, by performing exponential function fitting on the micro-environment pressure time series and continuously monitoring the changing trends of the goodness of fit and the exponential growth coefficient, it can accurately capture the irreversible exponential approach characteristics of micro-environment pressure from negative pressure to atmospheric pressure at the evolution endpoint of the formation of a macro-leakage channel through micro-damage to the oil seal lip. It generates a forced shutdown maintenance command before macro-leakage occurs, thereby effectively eliminating the bottleneck of sudden oil leakage shutdown and significantly improving the reliability and safety of critical fluid equipment in long-term operation scenarios.

[0119] Example 2

[0120] It should be further explained that this embodiment also provides a theoretical internal leakage rate drift model per unit time. Its core purpose is to determine the evolution of internal leakage rate of the skeleton oil seal from health to failure based on bench test calibration. This provides a quantitative verification basis for online determination of the three key failure stages of the oil seal: critical wear due to oil exhaust, local micromaterial spalling, and formation of through-channels. It can form multi-dimensional cross-validation with real-time collected microenvironment pressure, vibration and shock characteristic values, and operating condition parameters, effectively filtering out the risk of misjudgment caused by operating condition disturbances, sensor anomalies, and external vibration interference. At the same time, it realizes the quantitative assessment of the degree of oil seal deterioration and the prediction of leakage development trend. The preset method of the theoretical internal leakage rate drift model per unit time is as follows:

[0121] In bench tests, a micro gas flow meter or a condenser-type oil-gas concentration detection device was deployed at the outlet of the sealing cavity downstream of the oil seal. For typical operating conditions of the SGP series gear pump within the range of speed, operating pressure, and medium viscosity given in the sample, steady-state measurements were performed on the amount of oil-gas escaping through the oil seal or the internal leakage rate per unit time at four stages: when the oil seal is in a healthy state, the lip is entering the critical wear stage due to oil shortage, the lip is experiencing localized micromaterial peeling, and the lip's airtight contact zone forms a through-channel microchannel. The measured internal leakage rate values ​​under different operating conditions at each evolution stage were piecewise fitted according to speed, operating pressure difference, medium viscosity, and deterioration stage to construct a theoretical internal leakage rate drift model per unit time. The piecewise basis of this theoretical internal leakage rate drift model per unit time is the specific piecewise function of the four typical deterioration stages of the oil seal from a healthy state to failure, including:

[0122] In this embodiment, the fitting of the first stage, the healthy state stage, is specifically as follows:

[0123] Under healthy conditions, a complete hydrodynamic lubrication film is maintained between the oil seal lip and the journal, and theoretically, no measurable macroscopic internal leakage occurs; the theoretical internal leakage rate fitting function per unit time in this stage is a constant zero function, denoted as Q. th1 It equals zero. At this point, the microenvironment pressure remains stably maintained in the negative pressure zone below atmospheric pressure, without any trend of drift over time.

[0124] In this embodiment, the fitting of the second stage, the critical wear stage of the exhausted oil, is specifically as follows:

[0125] When the oil seal enters the critical wear stage due to insufficient oil, the lip lubricating film thins or partially ruptures, the contact area enters a boundary lubrication state, and tiny gaps begin to appear at the sealing interface, resulting in a very small amount of initial leakage. The internal leakage rate at this stage is exponentially related to the operating pressure difference, inversely proportional to the medium viscosity, and directly proportional to the rotational speed. This embodiment uses the following fitting function to piecewise fit the measured values ​​under each operating condition at this stage. Where ΔP is the operating pressure difference, ω is the rotational speed, η is the medium viscosity, A is the stage proportionality coefficient, u is the pressure difference index factor, and m is the viscosity index factor. The coefficients A and factors n and m are determined by those skilled in the art through least-squares nonlinear regression based on multi-condition measurement data for the corresponding stage.

[0126] In this embodiment, the fitting of the third stage, the local micromaterial peeling stage, is specifically as follows:

[0127] When localized micro-material peeling occurs at the oil seal lip, randomly distributed micro-pits or material spalling marks appear at the sealing interface. These defects form additional micro-leakage channels. The internal leakage rate at this stage is based on the oil-starved wear stage, superimposed with a fluctuation component related to the impact characteristic value. The fitting function used in this embodiment is as follows: Among them, I impact The impact characteristic value is extracted from the vibration envelope demodulation, and B and C are the fitting coefficients for this stage. This impact characteristic value is positively correlated with the degree of lip material spalling. The coefficients B and C are determined by those skilled in the art through linear regression based on the correlation coefficient between the vibration impact characteristic value of this stage and the corresponding internal leakage rate measurement value. The coefficients B and C for each working condition are pre-stored in the seal failure association database and correspond one-to-one with the speed and working pressure difference nodes.

[0128] In this embodiment, the fitting of the fourth stage, the through-channel formation stage, is specifically as follows:

[0129] Once a continuous microchannel forms in the airtight contact zone of the oil seal lip, a continuous leakage path appears at the sealing interface, penetrating both the inner and outer sides of the lip. The internal leakage rate increases exponentially and is no longer constrained by the state of the lubricating film. An exponential growth model is used to fit this stage. Q critical The initial leakage rate at the start of this stage is the estimated internal leakage rate at the end of the third stage. t represents the duration from the moment the microenvironment pressure drift slope triggers the irreversible exponential approximation judgment. k is the leakage development rate coefficient. The value of coefficient k is related to the current operating pressure difference and the medium viscosity, and is determined by those skilled in the art through experimental data fitting based on the growth slope of the internal leakage rate under different operating conditions in this stage. The coefficient k for each operating condition is pre-stored in the seal failure correlation database, corresponding one-to-one with the nodes of operating pressure difference and medium viscosity.

[0130] It should be further explained that the internal leakage rate-microenvironment pressure mapping relationship, which is calibrated simultaneously with the theoretical internal leakage rate drift model per unit time, is pre-set in the following way: during the calibration process of each working condition of the bench test, the steady-state value of the microenvironment pressure in the oil seal front cavity corresponding to each steady-state internal leakage rate measurement is recorded simultaneously. A linear mapping equation of "increase in leakage rate per unit time - increase in microenvironment pressure per unit time" is obtained through linear regression fitting. ,in G represents the increase in microenvironmental stress relative to a baseline value for healthy status, and G is the mapping proportionality coefficient. The G value is the increment of the theoretical internal leakage rate relative to the zero baseline of the healthy state. It is related to the speed and working pressure difference of the current operating condition and is pre-stored in the database. It enables the calculation of the corresponding micro-environment pressure baseline value through the theoretical internal leakage rate, or the deduction of the actual internal leakage rate range through the measured value of the micro-environment pressure.

[0131] As a specific example, in this embodiment, the entire leakage evolution process of the oil seal from health to failure is divided into four typical deterioration stages that can be automatically determined by online monitoring parameters: The first stage is the healthy state stage. Taking a certain type of SGP gear pump in steady-state operation at a speed of 1500 rpm, a working pressure difference of 20 bar, and a medium viscosity of 34 mm² / s as an example, the microenvironment pressure is stably maintained at -0.2 bar, which is within the safe range of -0.4 bar, the lower limit of the shaft seal inlet pressure allowed under this condition. At the same time, the vibration in the 2kHz to 10kHz frequency band is stable. The root mean square value of the dynamic impact characteristic is 0.05g, which is stably maintained below the baseline of 0.10g in the characteristic frequency band of this operating condition. In the second stage, the critical wear stage due to oil exhaust, when the suction side filter gradually becomes clogged, causing the microenvironment pressure to drop to -0.45bar, which is lower than the lower limit of the allowable shaft seal inlet pressure corresponding to the current operating condition (-0.4bar), and this state continues for a first preset duration of 5 seconds, the system generates the first trigger mark. At this time, the vibration impact characteristic value has not yet increased significantly. In the third stage, the local micromaterial peeling stage, when the microenvironment pressure has triggered the first trigger mark... Against this background, when the fluctuation range of the real-time effective differential pressure is within the preset stable threshold of 0.5 bar, the root mean square value of the vibration impact characteristic value gradually increases from the baseline of 0.10 g to 0.18 g, exceeding the baseline of the vibration intensity of the characteristic frequency band of this working condition. Moreover, the interval of the impact peak shows a non-integer multiple relationship with the frequency of 25 Hz, that is, the impact interval is non-uniform and non-periodic. The system generates a second trigger mark. In the fourth stage, the formation stage of the penetrating microchannel, within the second preset time of 10 minutes after the generation of the first or second trigger mark, the microenvironment pressure increases from -0. The pressure drops from 45 bar to zero in an exponential curve pattern, with the pressure values ​​sequentially decreasing to -0.43 bar, -0.38 bar, -0.30 bar, -0.19 bar, and -0.10 bar. The slope of the drift gradually increases over time. Once this irreversible exponential approach trend is continuously monitored and confirmed, the system determines that the microenvironment pressure has irreversibly recovered from a negative pressure state to atmospheric pressure. A through-channel has formed in the airtight contact zone of the oil seal lip, and the sealing interface is in the final critical state before macroscopic leakage. A forced shutdown maintenance command is then generated. It should be noted that the values ​​in this step are exemplary descriptions to facilitate understanding of the working principle of this solution. In actual implementation, the specific values ​​of parameters such as the lower limit of the shaft seal inlet pressure threshold curve, the baseline of the characteristic frequency band vibration intensity, the first preset duration, the second preset duration, and the preset stability threshold should be determined by those skilled in the art through bench testing based on the specifications of the target pump type, the model and material of the rotary oil seal, the operating conditions, and the actual operating conditions. This application does not impose specific limitations on these values.

[0132] It should be further explained that the specific implementation method of multi-dimensional cross-validation in this embodiment is as follows;

[0133] In this embodiment, the health status benchmark verification corresponds to the first stage of the model. The entire monitoring process is continuously executed and serves as the benchmark reference for all subsequent deterioration stages. It is continuously executed after the pump starts and enters steady-state operation. If there is no warning trigger, the verification is performed by default in a loop. The specific implementation method is as follows: Within each monitoring cycle, the current operating speed, medium viscosity, and real-time effective differential pressure are simultaneously collected as input conditions for model retrieval; using the above operating condition parameters as joint retrieval conditions, the first-stage fitting function of the theoretical internal leakage rate drift model per unit time, i.e., the constant zero function where the theoretical internal leakage rate equals zero, is retrieved from the seal failure association database; at the same time, the microenvironment pressure benchmark value and characteristic frequency band vibration intensity baseline corresponding to the oil seal health state under this operating condition are retrieved; the real-time collected microenvironment pressure measurement value and vibration impact characteristic value are compared with the retrieved health state benchmark value, and the matching of the theoretical internal leakage rate of the first stage of the model equaling zero with the measured operating condition is verified. When the microenvironment pressure is stable within the safe range of the health state and the vibration impact characteristic value is lower than the baseline, the theoretical calculation result is determined to be consistent with the measured state; when the verification is passed, the oil seal is determined to be in a healthy state, no warning flag is triggered, and the system continues to execute the benchmark verification of the next monitoring cycle; when the verification fails, the specific verification process of the corresponding deterioration stage is directly entered. As a specific example, taking the steady-state operating condition of the SGP080 specification oil gear pump in this embodiment as an example, the current operating speed is 1500 rpm, the working pressure difference is 20 bar, and the medium viscosity is 34 mm² / s. By jointly retrieving the corresponding operating condition, the healthy state microenvironment pressure benchmark value is -0.2 bar, the vibration intensity baseline is 0.10 g, the theoretical internal leakage rate of the first stage of the model is zero, the real-time collected microenvironment pressure is stably maintained at -0.2 bar, and the root mean square value of the vibration impact characteristic value is 0.05 g. The measured value is completely matched with the healthy benchmark value and the theoretical calculation result. The benchmark verification is passed, and the oil seal is determined to be in a healthy state.

[0134] In this embodiment, the model verification of the critical wear stage of the oil-depleted oil corresponds to the second stage of the model. As a necessary prerequisite for generating the first trigger mark, it is executed synchronously with the pressure threshold determination in S3. The first trigger mark can only be recorded when the pressure threshold determination and model verification pass simultaneously. Specifically, when the real-time collected microenvironment pressure is lower than the allowable shaft seal inlet pressure limit corresponding to the current operating condition, while starting the first preset timer, the current operating speed, medium viscosity, and real-time effective pressure difference are locked as fixed input parameters for model verification. Using the locked operating condition parameters as joint retrieval conditions, the fitting coefficients corresponding to the second stage of the theoretical internal leakage rate drift model per unit time are retrieved from the seal failure association database. These coefficients include the stage proportionality coefficient A, pressure difference index factor n, and viscosity index factor m, as well as the internal leakage rate and microenvironment pressure under this operating condition. Mapping coefficient K; Substitute the locked operating parameters into the second-stage fitting function to calculate the theoretical internal leakage rate of the critical wear stage of the oil-starved area under the current operating conditions; Substitute the theoretical internal leakage rate into the mapping relationship between the internal leakage rate and the micro-environment pressure to calculate the theoretical increase in micro-environment pressure corresponding to the theoretical internal leakage rate; Compare the measured micro-environment pressure change trend with the theoretical increase in micro-environment pressure within the first preset time period. When the two change trends are consistent and the relative deviation is less than or equal to the preset threshold, the verification is passed, and the oil seal lip is determined to have entered the critical wear stage of the oil-starved area, allowing the recording of the first trigger. In this embodiment, the preset threshold is set as follows: During the bench calibration stage before system commissioning, those skilled in the art, under the condition that the oil seal is in a healthy state, simulate typical operating condition disturbances such as gradual blockage of the suction pipeline filter and micro-movement of the suction side valve at the steady-state operating points of each speed range and working pressure difference combination covered by the sealing failure association database. Simultaneously, the measured change in micro-environment pressure during each disturbance process and the theoretical increase calculated by the theoretical internal leakage rate drift model and the internal leakage rate-micro-environment pressure mapping relationship under the corresponding operating conditions are collected. The relative deviation between the measured value and the theoretical value under all operating condition nodes is statistically analyzed, and the mean and standard deviation of the relative deviation are calculated. The mean plus three times the standard deviation is taken as the statistical upper limit, and a preset engineering margin is added on this basis. The resulting value is determined as the preset threshold and stored in the sealing failure association database for direct retrieval during the verification of the oil-deficient critical wear model in the online monitoring stage. Thus, under the premise of ensuring sufficient identification sensitivity for the real oil-deficient critical wear state, the probability of model verification misjudgment caused by normal fluctuations in operating conditions is controlled within a statistically acceptable range.

[0135] When the trends of the two are opposite or the relative deviation is greater than the preset threshold, the verification fails, and it is determined to be a sensor malfunction or a momentary disturbance in the operating condition. The first trigger flag is not triggered, and a sensor verification prompt is generated. As a specific example, taking the triggering condition of the SGP080 specification oil gear pump in this embodiment as an example, the current operating speed is 1750 rpm, the medium viscosity is 34 mm² / s, the real-time effective differential pressure is 20.45 bar, and the measured value of the micro-environment pressure is -0.45 bar, which is lower than the lower limit of the allowable shaft seal inlet pressure for the corresponding operating condition, -0.39 bar. While starting the first preset duration of 5 seconds, the above operating condition parameters are locked, and the corresponding stage proportional coefficient A, differential pressure index factor n, and viscosity index factor m are retrieved through joint retrieval. The mapping coefficient K was substituted into the model to calculate a theoretical internal leakage rate of approximately 0.12 mL / min, corresponding to a theoretical increase in microenvironmental pressure of approximately 0.03 bar. Within a 5-second timing period, the measured microenvironmental pressure continuously rose from -0.45 bar towards atmospheric pressure, with a cumulative increase of 0.028 bar. The relative deviation from the theoretical increase was approximately 6.7%, which was less than the preset threshold. The trend of change was completely consistent with the model verification, and the model was successfully verified. It matched the pressure threshold determination result of step S3 in the original scheme. The first trigger mark was recorded simultaneously, and a first-level early warning command was generated.

[0136] In this embodiment, the model verification of the local micromaterial peeling stage corresponds to the third stage of the model. As a necessary prerequisite for generating the second trigger marker, it is executed synchronously with the vibration characteristic three-condition judgment in S4. The second trigger marker can only be recorded when the three-condition judgment and model verification pass simultaneously. Specifically, when the real-time calculated impact characteristic value exceeds the vibration intensity baseline corresponding to the current operating condition, the current operating speed, medium viscosity, real-time effective pressure difference, and current impact characteristic value are locked as fixed input parameters for model verification. Using the locked operating condition parameters as joint retrieval conditions, the fitting coefficients corresponding to the third stage of the theoretical internal leakage rate drift model per unit time are retrieved from the sealing failure association database. These coefficients include the stage proportionality coefficient B, impact characteristic value coefficient C, pressure difference index factor n, viscosity index factor m, and the mapping coefficient K between the internal leakage rate and microenvironment pressure under this operating condition. The locked operating condition parameters and impact characteristic values ​​are then substituted... The third-stage fitting function is used to calculate the theoretical internal leakage rate of the local micromaterial spalling stage under the current operating condition. By substituting the theoretical internal leakage rate into the mapping relationship between the internal leakage rate and the microenvironment pressure, the theoretical cumulative increase in microenvironment pressure corresponding to this theoretical internal leakage rate is calculated. When the three conditions of S4 impact exceeding the baseline, coefficient of variation exceeding the non-periodic threshold, and stable operating condition are simultaneously met, model verification is performed. The measured cumulative increase in microenvironment pressure is compared with the theoretically calculated value. When the relative deviation between the two is less than or equal to the preset allowable range, the verification is passed, and it is determined that local micromaterial spalling has occurred at the oil seal lip, allowing the recording of the second trigger mark. Note: In this embodiment, the preset allowable range is preset in the following way: During the bench calibration stage before system commissioning, those skilled in the art simulate local micro-material spalling at the oil seal lip by artificially creating micro-spalling defects. At steady-state operating points in each speed range and working pressure difference combination covered by the seal failure correlation database, following the same sensor placement, signal acquisition parameters, and envelope demodulation processing procedure as the online monitoring stage, the actual cumulative increase in micro-environment pressure under each operating condition is simultaneously collected, along with the fitting function of the third stage of the theoretical internal leakage rate drift model per unit time and the mapping relationship between internal leakage rate and micro-environment pressure. The theoretical cumulative increase is calculated; the relative deviation between the measured values ​​and the theoretical values ​​under all working conditions is statistically analyzed, the mean and standard deviation of the relative deviation are calculated, the mean plus three times the standard deviation is taken as the statistical upper limit, and a preset engineering margin is added on this basis. The resulting value is determined as the preset allowable range and stored in the sealing failure association database for direct retrieval during the verification of the local microscopic material spalling model in the online monitoring stage. Thus, while ensuring sufficient sensitivity to identify the real material spalling state, the probability of verification misjudgment caused by model fitting error and normal fluctuations in working conditions is controlled within a statistically acceptable range.

[0137] When the relative deviation between the two exceeds the preset allowable range, the verification fails, and it is determined to be external vibration interference, and the second trigger flag is not triggered. As a specific example, taking the triggering condition of the SGP080 specification oil gear pump in this embodiment as an example, the current operating speed is 1500 rpm, the medium viscosity is 34 mm² / s, the real-time effective pressure difference is 15 bar, and the measured impact characteristic value is 0.18 g, which exceeds the corresponding vibration intensity baseline of 0.10 g. The above operating condition parameters are locked, and the corresponding stage proportional coefficient B, impact characteristic value coefficient C, pressure difference index factor n, viscosity index factor m, and mapping coefficient K are retrieved by joint retrieval. Substituting them into the model, the theoretical internal leakage rate is calculated to be approximately 0.35 mL / min, corresponding to a microenvironment pressure of 0.35 mL / min. The theoretical cumulative increase in force is approximately 0.0875 bar. The original scheme's S4 step simultaneously determined that the impact interval variation coefficient of 0.90 is greater than the preset threshold of 0.35, and the real-time effective pressure difference fluctuation amplitude of 0.2 bar is less than the preset stability threshold of 0.5 bar. All three conditions are met simultaneously. The measured microenvironment pressure has increased by 0.09 bar from the healthy baseline value of -0.2 bar to the current moment, which is about 2.9% higher than the theoretical cumulative increase. This is less than the preset allowable range, so the model verification is successful and matches the determination result of the original scheme's S4 step. The second trigger mark is recorded simultaneously, and a secondary warning instruction is generated.

[0138] In this embodiment, the model verification of the through-hole microchannel formation stage corresponds to the fourth stage of the model. As a necessary prerequisite for generating the forced shutdown maintenance command, it is executed synchronously with the microenvironment pressure exponential approximation judgment in S5. The forced shutdown maintenance command can only be generated when the exponential fitting judgment and model verification pass simultaneously. Specifically, after the first or second trigger flag is generated, while starting the second preset duration monitoring window, the rotational speed, working pressure difference, and medium viscosity of the current operating condition are locked. The theoretical internal leakage rate at the end of the third stage is retrieved as the initial leakage rate of the through-hole microchannel stage. Using the locked operating condition parameters as joint retrieval conditions, the leakage development rate coefficient k corresponding to the fourth stage of the theoretical internal leakage rate drift model per unit time, and the mapping coefficient K between the internal leakage rate and the microenvironment pressure under this operating condition are retrieved from the sealing failure association database. Within the monitoring window, with the sampling period as the step size, the fourth stage exponential growth model is continuously substituted to calculate the theoretical internal leakage rate within the entire monitoring window. Leakage rate time series; by mapping the internal leakage rate to the microenvironment pressure, the theoretical internal leakage rate time series is converted into the corresponding theoretical microenvironment pressure time series; the goodness of fit between the measured microenvironment pressure time series and the theoretical microenvironment pressure time series within the monitoring window is calculated. When the coefficient of determination R² is greater than the preset goodness of fit threshold, and the measured pressure continues to irreversibly approach atmospheric pressure without any downward trend, the verification is passed, and it is determined that the airtight contact zone of the oil seal lip has formed a through-channel microchannel, allowing the generation of a forced shutdown maintenance command; if the goodness of fit does not meet the threshold, or the measured pressure shows a downward trend, the verification fails, and it is determined to be a normal operating condition fluctuation, and no shutdown command is generated. As a specific example, taking the triggering condition of the SGP080 specification oil gear pump in this embodiment as an example, the second trigger marker has been generated, and the second preset monitoring window of 10 minutes is started. The current operating conditions are locked at a speed of 1500 rpm, a working pressure difference of 20 bar, and a medium viscosity of 34 mm² / s. The theoretical internal leakage rate at the end of the third stage is retrieved as the initial leakage rate. The leakage development rate coefficient k and the mapping coefficient K under the corresponding operating conditions are retrieved through joint retrieval. Within the monitoring window, with a sampling period of 100 milliseconds, the theoretical internal leakage rate time series is continuously substituted into the model calculation. The measured microenvironment pressure time series was simultaneously converted into a theoretical microenvironment pressure time series. The measured microenvironment pressure time series were -0.45 bar, -0.43 bar, -0.38 bar, -0.30 bar, -0.19 bar, and -0.10 bar, continuously rising towards zero bar without any downward trend. After fitting calculation, the coefficient of determination R² between the measured pressure time series and the theoretical pressure time series was 0.97, which is greater than the preset threshold of 0.9. The model verification passed and matched the exponential approximation judgment result of step S5 of the original scheme. A forced shutdown maintenance command was immediately generated.

[0139] Example 3

[0140] Please see Figure 3 Another embodiment of the present invention provides: an online monitoring and leakage identification system for the sealing status of rotating machinery, comprising:

[0141] The database preset module is configured as follows: a preset database of pump operating characteristic parameters and seal failure association, the database including at least the allowable shaft seal inlet pressure threshold curve under different speed ranges, working pressure difference and medium viscosity combinations, the theoretical internal leakage rate drift model per unit time and the characteristic frequency band vibration intensity baseline of preset measuring points of the shell.

[0142] The acquisition module is configured to: acquire in real time the working pressure of the pump outlet pipeline, the micro-environmental pressure of the chamber between the oil seal and the front bearing, and the vibration time-domain waveform of the pump body at the oil seal seat end face through a pressure transmitter deployed on the pump outlet pipeline, a micro differential pressure sensor deployed near the oil seal mounting stop of the pump body, and a broadband vibration sensor attached to the end face of the rear cover or oil seal seat.

[0143] The critical wear determination module is configured to: calculate the real-time effective pressure difference based on the working pressure and the micro-environment pressure, and obtain the corresponding allowable shaft seal inlet pressure threshold curve based on the current operating speed and medium viscosity; when the micro-environment pressure is lower than the lower limit of the allowable shaft seal inlet pressure threshold curve and continues for a first preset time, the oil seal lip lubrication state is determined to have entered the oil shortage critical wear stage, the first trigger mark is recorded and a first-level warning command is generated to prompt adjustment of the suction pipeline conditions;

[0144] The spalling identification module is configured to: perform envelope demodulation processing on the vibration time-domain waveform, extract repetitive impact characteristic values ​​of a preset frequency band, and when the impact characteristic value exceeds the vibration intensity baseline of the characteristic frequency band and shows an increasing trend that is not an integer multiple of the current frequency, while the fluctuation amplitude of the real-time effective pressure difference is within a preset stable threshold range, it is determined that local micro-material spalling has occurred at the oil seal lip, a second trigger mark is recorded, and a secondary warning command is generated; the preset frequency band is 2kHz to 10kHz.

[0145] The decision module is configured to: continuously calculate the time drift slope of the microenvironment pressure within a second preset time period after the first trigger mark or the second trigger mark is generated; if the drift slope indicates that the microenvironment pressure is irreversibly exponentially approaching atmospheric pressure from a negative pressure state, then it is determined that the airtight contact zone of the oil seal lip has formed a through-channel microchannel, the sealing interface is in a pre-failure state evolving from micro wear to macro leakage, and a forced shutdown maintenance command is generated.

[0146] The embodiments of the present invention have been described above with reference to the accompanying drawings. However, the present invention is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments under the guidance of the present invention without departing from the spirit and scope of the claims. All of these variations are within the protection scope of the present invention.

[0147] If the technical solution disclosed herein involves personal information, the product using this technical solution has clearly informed the user of the personal information processing rules and obtained the user's voluntary consent before processing the personal information. If the technical solution disclosed herein involves sensitive personal information, the product using this technical solution has obtained the user's separate consent before processing the sensitive personal information, and also meets the requirement of "express consent". For example, at personal information collection devices such as cameras, clear and prominent signs are set up to inform users that they have entered the scope of personal information collection and that personal information will be collected. If an individual voluntarily enters the collection scope, it is deemed that they have agreed to the collection of their personal information; or on the personal information processing device, with clear signs / information informing users of the personal information processing rules, authorization is obtained from the individual through pop-up information or by asking the individual to upload their personal information; wherein, the personal information processing rules may include information such as the personal information processor, the purpose of personal information processing, the processing method, and the types of personal information processed.

Claims

1. A method for online monitoring of sealing status and leakage identification of rotating machinery, applied to a fluid transport system including an oil delivery gear pump, wherein a skeleton rotary oil seal is installed between the rotating shaft and the housing of the gear pump, characterized in that, include: The working pressure of the pump outlet pipeline, the microenvironment pressure of the chamber between the oil seal and the front bearing, and the time-domain waveform of the pump body vibration at the end face of the oil seal seat are obtained. The real-time effective pressure difference is calculated based on the working pressure and the micro-environment pressure, and a preset allowable shaft seal inlet pressure threshold curve is obtained based on the current operating speed and medium viscosity. When the micro-environment pressure is lower than the lower limit of the allowable shaft seal inlet pressure threshold curve and continues for a first preset time, the oil seal lip lubrication state is determined to have entered the oil shortage critical wear stage, and a first trigger mark is recorded. Envelope demodulation is performed on the vibration time-domain waveform to extract the repetitive impact characteristic value of the preset frequency band. When the impact characteristic value exceeds the corresponding vibration intensity baseline and shows an increasing trend that is not an integer multiple of the current frequency, and the real-time effective differential pressure display shows stable operating conditions, it is determined that local micro-material peeling has occurred at the oil seal lip, and a second trigger mark is recorded. Within a second preset time period after the first or second trigger mark is generated, the time drift slope of the microenvironment pressure is continuously calculated. If the drift slope indicates that the microenvironment pressure is irreversibly and exponentially approaching atmospheric pressure from a negative pressure state, it is determined that the airtight contact zone of the oil seal lip has formed a through-channel microchannel, and the sealing interface is in a state before failure, evolving from micro-wear to macro-leakage, and a forced shutdown maintenance command is generated.

2. The method for online monitoring of sealing status and leakage identification of rotating machinery as described in claim 1, characterized in that, The allowable shaft seal inlet pressure threshold curve and the vibration intensity baseline are stored in a seal failure correlation database containing different speed ranges, working pressure differences and media viscosity combinations. The database also stores a theoretical internal leakage rate drift model per unit time. When the first trigger mark is recorded, a first-level warning instruction is also generated to prompt adjustment of the inhalation tubing conditions; when the second trigger mark is recorded, a second-level warning instruction is also generated. The preset frequency band is from 2kHz to 10kHz, the vibration intensity baseline is the characteristic frequency band vibration intensity baseline of the preset measuring point of the shell, and the operating condition is stable, specifically, the fluctuation amplitude of the real-time effective pressure difference is within the preset stable threshold range.

3. The method for online monitoring of sealing status and leakage identification of rotating machinery as described in claim 2, characterized in that, The determination that the oil seal lip has entered the critical wear stage due to insufficient oil lubrication is specifically as follows: Based on the real-time acquired current operating speed and current medium viscosity, a joint search is performed in a preset seal failure association database, and a dynamic allowable lower limit value of shaft seal inlet pressure matching the current operating conditions is obtained through two-dimensional interpolation calculation; wherein, the seal failure association database stores allowable lower limit values ​​of shaft seal inlet pressure under different speed ranges and different medium viscosity combinations, which are pre-calibrated through bench tests; The real-time collected microenvironment pressure value is compared with the lower limit of the dynamically permissible shaft seal inlet pressure. When the microenvironment pressure value is lower than the lower limit of the dynamic allowable shaft seal inlet pressure, the first preset duration timer is started. If the microenvironment pressure value rises to a level greater than or equal to the lower limit of the dynamic allowable shaft seal inlet pressure within the first preset time period, it is determined that the microenvironment pressure value collected at the corresponding time exceeds the limit due to instantaneous fluctuations in the working condition. The first preset time period is then reset to zero, and the timing is restarted when the microenvironment pressure value falls below the lower limit of the dynamic allowable shaft seal inlet pressure again. When the cumulative duration for which the microenvironment pressure value is continuously lower than the lower limit of the dynamic allowable shaft seal inlet pressure reaches the first preset duration, it is determined that the oil seal lip has entered the critical wear stage due to the continuous thinning of the lubricating film, and the first trigger mark is recorded.

4. The method for online monitoring of sealing status and leakage identification of rotating machinery as described in claim 3, characterized in that, The envelope demodulation processing of the vibration time-domain waveform to extract repetitive impact characteristic values ​​of a preset frequency band includes: The vibration time-domain waveform acquired by the broadband vibration sensor attached to the end face of the rear cover or oil seal seat is subjected to bandpass filtering with a passband frequency of 2kHz to 10kHz to obtain the filtered vibration signal. The 2kHz to 10kHz passband covers the resonant response frequency band excited by the impact of the peeling of the micromaterial of the oil seal lip on the pump body structure. The filtered vibration signal is subjected to Hilbert transform, and the amplitude envelope of the analytical signal is calculated to obtain the impact pulse envelope signal; For the impact pulse envelope signal, peak detection is performed within a preset analysis time window. Local peak points with amplitudes exceeding a preset trigger level are identified as impact events, and the timestamps and amplitudes of each impact event are recorded.

5. The method for online monitoring of sealing status and leakage identification of rotating machinery as described in claim 4, characterized in that, The process of envelope demodulating the vibration time-domain waveform and extracting repetitive impact characteristic values ​​of a preset frequency band further includes: Calculate the root mean square value of the amplitude of all impact events within the analysis time window, and use it as the current impact characteristic value; The current impact characteristic value is compared with the characteristic frequency band vibration intensity baseline stored in the seal failure association database. The characteristic frequency band vibration intensity baseline is the root mean square value of the impact characteristic value measured under the current speed and current working pressure difference combination conditions when the oil seal is in a healthy state. When the current impact characteristic value is greater than the characteristic frequency band vibration intensity baseline, the coefficient of variation of the time interval between each adjacent impact event within the analysis time window is calculated. The coefficient of variation is the ratio of the standard deviation of the time interval to the mean of the time interval.

6. The method for online monitoring of sealing status and leakage identification of rotating machinery as described in claim 5, characterized in that, The process of envelope demodulating the vibration time-domain waveform and extracting repetitive impact characteristic values ​​of a preset frequency band further includes: When the coefficient of variation is greater than a preset non-periodic threshold, the time interval of the impact event is determined to be unevenly distributed, and the increase of the current impact characteristic value relative to the vibration intensity baseline is not an integer multiple of the current frequency. Simultaneously, the real-time effective pressure difference calculated based on the pump outlet working pressure and the micro-environment pressure is obtained. When the fluctuation range of the real-time effective pressure difference is within the preset stable threshold range, the current operating condition is determined to be stable. When the current impact characteristic value is greater than the vibration intensity baseline, the coefficient of variation is greater than the non-periodic threshold, and the fluctuation amplitude is within the preset stability threshold range, it is determined that local micromaterial peeling has occurred at the oil seal lip, and a second trigger mark is recorded.

7. The method for online monitoring of sealing status and leakage identification of rotating machinery as described in claim 6, characterized in that, Calculate the amplitude envelope of the analytic signal to obtain the impulse pulse envelope signal, specifically including: The filtered vibration signal obtained after the bandpass filtering is processed by the frequency domain Hilbert transform algorithm based on the fast Fourier transform to obtain an orthogonal signal that is 90 degrees out of phase with the filtered vibration signal. The discrete sampling sequence of the filtered vibration signal is taken as the real part sequence, and the orthogonal signal is taken as the imaginary part sequence. An analytical signal is constructed point by point. The value of the analytical signal at the nth sampling point is the sum of the nth value of the real part sequence and the nth value of the imaginary part sequence multiplied by the imaginary unit j. The instantaneous amplitude is calculated point by point for the analytical signal. The instantaneous amplitude at the nth sampling point is equal to the square root of the sum of the square of the nth value of the real part sequence and the square of the nth value of the imaginary part sequence. The instantaneous amplitude sequence, which is composed of the instantaneous amplitudes of all sampling points arranged in chronological order, is used as the impulse pulse envelope signal.

8. The method for online monitoring of sealing status and leakage identification of rotating machinery as described in claim 7, characterized in that, The determination that the airtight contact zone of the oil seal lip has formed a through-channel includes: After the first trigger flag or the second trigger flag is generated, the second preset duration timer is started, and within the second preset duration, the microenvironment pressure value is continuously acquired at a preset sampling period. The microenvironment pressure values ​​arranged in the order of sampling time constitute the microenvironment pressure time series. The microenvironment pressure time series was fitted with an exponential function, the fitting function being P_seal(t)=a×e (b×t) +c, where t is the time variable with the start time of the second preset duration as zero, P_seal(t) is the microenvironment pressure fitting value at time t, a, b, and c are fitting coefficients determined by the least squares method, b is the exponential growth coefficient characterizing the microenvironment pressure drift rate, and c is the asymptotic value of the fitting function in the direction of atmospheric pressure. Calculate the goodness of fit R² of the exponential function and the exponential growth coefficient b. When the goodness of fit R² of the exponential function is greater than the preset goodness of fit threshold and the exponential growth coefficient b is positive, it is determined that the microenvironment pressure time series shows an irreversible drift trend approaching zero pressure value at an exponential rate.

9. The method for online monitoring of sealing status and leakage identification of rotating machinery as described in claim 8, characterized in that, The determination that the airtight contact zone of the oil seal lip has formed a through-channel microchannel also includes: If, before the end of the second preset time period, the microenvironment pressure time series shows an irreversible drift trend approaching zero pressure value at an exponential rate, it is continuously monitored and confirmed that the oil seal lip airtight contact zone has formed a through-channel microchannel and the sealing interface is in a state of failure before evolving from micro wear to macro leakage, and a forced shutdown maintenance command is generated. If, at the end of the second preset duration, the goodness of fit R² of the exponential function is less than or equal to the preset goodness of fit threshold, or the exponential growth coefficient b is negative or zero, then it is determined that the change in the microenvironment pressure still belongs to normal operating condition fluctuations or that the oil seal deterioration has not yet entered the leakage evolution endpoint stage. The second preset duration timer that has been started is cleared, and the online monitoring process continues to be executed.

10. A system for online monitoring of sealing condition and leakage identification of rotating machinery, used to implement the method for online monitoring of sealing condition and leakage identification of rotating machinery as described in any one of claims 1-9, characterized in that, include: The acquisition module is configured to acquire the working pressure of the pump outlet pipeline, the microenvironmental pressure of the chamber between the oil seal and the front bearing, and the time-domain waveform of the pump body vibration at the end face of the oil seal seat. The critical wear determination module is configured to: calculate the real-time effective pressure difference based on the working pressure and the micro-environment pressure, and obtain a preset allowable shaft seal inlet pressure threshold curve based on the current operating speed and medium viscosity; when the micro-environment pressure is lower than the lower limit of the allowable shaft seal inlet pressure threshold curve and continues for a first preset time, determine that the oil seal lip lubrication state has entered the oil shortage critical wear stage, and record the first trigger mark. The peeling identification module is configured to: perform envelope demodulation processing on the vibration time-domain waveform, extract the repetitive impact characteristic value of the preset frequency band, and when the impact characteristic value exceeds the corresponding vibration intensity baseline and shows an increasing trend that is not an integer multiple of the current frequency, and the real-time effective differential pressure display shows stable operating conditions, determine that local micro-material peeling has occurred at the oil seal lip, and record the second trigger mark. The decision module is configured to: continuously calculate the time drift slope of the microenvironment pressure within a second preset time period after the first trigger mark or the second trigger mark is generated; if the drift slope indicates that the microenvironment pressure is irreversibly exponentially approaching atmospheric pressure from a negative pressure state, then it is determined that the airtight contact zone of the oil seal lip has formed a through-channel microchannel, the sealing interface is in a pre-failure state evolving from micro wear to macro leakage, and a forced shutdown maintenance command is generated.

Images