Sensor-based bearing grease injection flow real-time monitoring method and system

Abstract

The invention provides a bearing grease injection flow real-time monitoring method and system based on a sensor, and the method comprises the steps: obtaining a real-time pressure sequence, a temperature value and an acoustic signal sequence of a grease injection pipeline, calculating a wall surface adhesion historical factor, and obtaining the real-time pressure sequence, the temperature value and the wall surface adhesion historical factor through the pressure change rate, the temperature value and the wall surface adhesion historical factor. A rheological compensation coefficient is calculated through a multivariable nonlinear mapping model, a reference grease injection flow is obtained by combining a real-time pressure sequence, the rheological compensation coefficient and a slip compensation coefficient, and for an acoustic signal sequence, energy attenuation is normalized by calculating transmission time delay and energy attenuation between adjacent sensors and utilizing the rheological compensation coefficient, so that the real-time pressure sequence is obtained. The matrix is compared with a preset bubble characteristic mode; if the similarity exceeds a threshold value, a bubble event is identified, the volume is estimated, the reference grease injection flow is corrected in real time, the flow corresponding to the bubble volume is deducted, and the real-time grease injection flow is obtained.

Description

Technical Field

[0001] This application belongs to the field of monitoring, and in particular relates to a sensor-based method and system for real-time monitoring of bearing grease injection flow. Background Technology

[0002] In heavy equipment sectors such as wind power, rail transportation, and metallurgy, bearings are core transmission components, and proper lubrication is essential for maintaining their optimal condition. Grease, a high-viscosity, non-Newtonian fluid, presents a significant technical challenge due to its complex properties, making accurate flow measurement a major technical hurdle in the industry. Existing grease flow monitoring technologies include volumetric methods based on piston displacement and direct measurement methods based on mass flow meters. Volumetric methods estimate grease volume by measuring pump stroke or piston displacement, but cannot provide real-time monitoring of the grease injection process and are ill-suited for detecting pipe blockages, leaks, and other anomalies. While mass flow meters offer higher accuracy, they are expensive, bulky, and poorly adapted to high-viscosity media like grease that easily contain solid particles, leading to clogging, measurement drift, and high maintenance costs. Furthermore, the rheological properties of grease change with temperature and pressure, and wall slippage occurs at pipe walls, causing errors in pressure-flow estimation methods based on fixed fluid models, which cannot be reliably compensated for. During storage, pumping, and transportation, grease is prone to air contamination, forming bubbles of varying sizes. Air bubbles can cause flow meter readings to be falsely high, resulting in the actual amount of grease injected into the bearing being far lower than the set value. This leads to false or insufficient lubrication of the bearing, posing a safety hazard to the equipment. Current technologies struggle to reliably identify, estimate the volume of air bubbles entrained in highly viscous media online and correct the flow rate, resulting in a long-term lack of accuracy and reliability in grease injection, failing to meet the precision lubrication requirements of high-end equipment. Summary of the Invention

[0003] This invention proposes a sensor-based real-time monitoring method for bearing grease injection flow rate, which addresses the problem that existing technologies struggle to reliably identify and estimate the volume of air bubbles entrained in highly viscous media online, thus failing to accurately correct flow rates. The method includes the following steps:

[0004] The real-time pressure sequence of the pressure sensor installed in the grease injection line, the real-time temperature sequence of the temperature sensor, and the acoustic signal sequence of the acoustic sensor array arranged along the axial direction of the grease injection line are obtained.

[0005] Based on the grease injection interval duration and the estimated volume of bubbles identified in the previous cycle, the wall adhesion history factor is calculated, and the slip compensation coefficient is determined based on the wall adhesion history factor; based on the pressure change rate of the real-time pressure sequence, the real-time temperature value, and the wall adhesion history factor, the rheological compensation coefficient is calculated through a preset multivariate nonlinear mapping model; combined with the real-time pressure sequence, the rheological compensation coefficient, and the slip compensation coefficient, the baseline grease injection flow rate is calculated;

[0006] Calculate the transmission time delay and energy attenuation of acoustic signals between adjacent sensors; normalize the energy attenuation using the rheological compensation coefficient, and perform time-domain correction on the transmission time delay using the reference grease injection flow rate; construct a feature matrix based on the normalized energy attenuation and the corrected transmission time delay; compare the similarity of the feature matrix with the bubble feature patterns in the preset pattern library, and if the similarity is greater than a preset threshold, identify it as a bubble event and estimate its volume;

[0007] The baseline grease injection flow rate is corrected in real time, and the flow rate corresponding to the estimated volume of the bubble event is deducted to obtain the real-time grease injection flow rate.

[0008] Optionally, the step of calculating the wall adhesion history factor based on the grease injection interval duration and the estimated volume of bubbles identified in the previous cycle, and determining the slip compensation coefficient based on the wall adhesion history factor, includes:

[0009] The wall adhesion history factor is obtained by weighting the grease injection interval duration and the estimated volume of the bubbles identified in the previous cycle.

[0010] Based on the wall adhesion history factor, the slip compensation coefficient is determined through a preset linear relationship.

[0011] Optionally, the step of calculating the rheological compensation coefficient based on the pressure change rate of the real-time pressure sequence, the real-time temperature value, and the wall adhesion history factor using a preset multivariate nonlinear mapping model includes:

[0012] The multivariate nonlinear mapping model is a feedforward neural network. The inputs of the network are the pressure change rate of the real-time pressure sequence, the real-time temperature value, and the wall adhesion history factor, and the output is the rheological compensation coefficient.

[0013] Optionally, calculating the baseline grease injection flow rate by combining the real-time pressure sequence, the rheological compensation coefficient, and the slip compensation coefficient includes:

[0014] The pressure difference ΔP between the inlet and outlet of the grease injection line is calculated based on the real-time pressure sequence, and combined with the line diameter D, line length L, grease base viscosity μ, and the rheological compensation coefficient. and the slip compensation coefficient The baseline grease injection flow rate is calculated according to the following formula. : .

[0015] Optionally, the calculation of the transmission time delay and energy attenuation of the acoustic signal between adjacent sensors includes:

[0016] The transmission time delay between adjacent sensor signals in the acoustic signal sequence is calculated using the generalized cross-correlation-phase transform method.

[0017] The energy attenuation of the signal is calculated based on the root mean square energy of the acoustic signals from the adjacent sensors.

[0018] Optionally, the step of normalizing the energy attenuation using the rheological compensation coefficient and performing time-domain correction of the transmission time delay using the reference grease injection flow rate includes:

[0019] The calculated energy attenuation is normalized using the rheological compensation coefficient.

[0020] The theoretical transmission time of the acoustic signal is calculated based on the reference grease injection flow rate, and the transmission time delay is corrected according to the difference between the theoretical transmission time and the actual calculated transmission time delay.

[0021] Optionally, the step of comparing the feature matrix with bubble feature patterns in a preset pattern library for similarity, and identifying it as a bubble event and estimating its volume if the similarity is greater than a preset threshold, includes:

[0022] The cosine similarity algorithm is used to calculate the similarity between the feature matrix and each bubble feature pattern in the preset pattern library;

[0023] If the similarity is greater than a preset threshold, it is identified as a bubble event, and the estimated volume of the bubble event is determined based on the energy decay characteristics corresponding to the matched bubble feature pattern and through a preset volume-energy decay mapping relationship.

[0024] Optionally, the step of real-time correction of the baseline grease injection flow rate, subtracting the flow rate corresponding to the estimated volume of the bubble event, to obtain the real-time grease injection flow rate includes:

[0025] When a bubble event is identified, the instantaneous flow rate corresponding to the bubble is calculated based on the estimated volume and the transit time of the bubble between the acoustic sensor arrays.

[0026] During the transit time of the bubble, the instantaneous flow rate corresponding to the bubble is subtracted from the baseline grease injection flow rate to obtain the real-time grease injection flow rate.

[0027] Furthermore, the present invention also relates to a sensor-based real-time monitoring system for bearing grease injection flow, comprising the following modules:

[0028] The acquisition module is used to acquire the real-time pressure sequence of the pressure sensor installed in the grease injection line, the real-time temperature sequence of the temperature sensor, and the acoustic signal sequence of the acoustic sensor array arranged along the axial direction of the grease injection line.

[0029] The calculation module is used to calculate the wall adhesion history factor based on the grease injection interval duration and the estimated volume of bubbles identified in the previous cycle, and to determine the slip compensation coefficient based on the wall adhesion history factor; to calculate the rheological compensation coefficient based on the pressure change rate of the real-time pressure sequence, the real-time temperature value, and the wall adhesion history factor through a preset multivariate nonlinear mapping model; and to calculate the baseline grease injection flow rate by combining the real-time pressure sequence, the rheological compensation coefficient, and the slip compensation coefficient.

[0030] The identification module is used to calculate the transmission time delay and energy attenuation of acoustic signals between adjacent sensors; normalize the energy attenuation using the rheological compensation coefficient, and perform time-domain correction on the transmission time delay using the reference grease injection flow rate; construct a feature matrix based on the normalized energy attenuation and the corrected transmission time delay; compare the similarity of the feature matrix with the bubble feature patterns in the preset pattern library, and if the similarity is greater than a preset threshold, it is identified as a bubble event and the volume is estimated.

[0031] The correction module is used to correct the baseline grease injection flow rate in real time, deduct the flow rate corresponding to the estimated volume of the bubble event, and obtain the real-time grease injection flow rate.

[0032] Preferably, the step of calculating the wall adhesion history factor based on the grease injection interval duration and the estimated volume of air bubbles identified in the previous cycle, and determining the slip compensation coefficient based on the wall adhesion history factor, includes:

[0033] The wall adhesion history factor is obtained by weighting the grease injection interval duration and the estimated volume of the bubbles identified in the previous cycle.

[0034] Based on the wall adhesion history factor, the slip compensation coefficient is determined through a preset linear relationship.

[0035] Preferably, the step of calculating the rheological compensation coefficient based on the pressure change rate of the real-time pressure sequence, the real-time temperature value, and the wall adhesion history factor using a preset multivariate nonlinear mapping model includes:

[0036] The multivariate nonlinear mapping model is a feedforward neural network. The inputs of the network are the pressure change rate of the real-time pressure sequence, the real-time temperature value, and the wall adhesion history factor, and the output is the rheological compensation coefficient.

[0037] Preferably, the step of calculating the baseline grease injection flow rate by combining the real-time pressure sequence, the rheological compensation coefficient, and the slip compensation coefficient includes:

[0038] The pressure difference ΔP between the inlet and outlet of the grease injection line is calculated based on the real-time pressure sequence, and combined with the line diameter D, line length L, grease base viscosity μ, and the rheological compensation coefficient. and the slip compensation coefficient The baseline grease injection flow rate is calculated according to the following formula. : .

[0039] Preferably, the calculation of the transmission time delay and energy attenuation of the acoustic signal between adjacent sensors includes:

[0040] The transmission time delay between adjacent sensor signals in the acoustic signal sequence is calculated using the generalized cross-correlation-phase transform method.

[0041] The energy attenuation of the signal is calculated based on the root mean square energy of the acoustic signals from the adjacent sensors.

[0042] Preferably, the step of normalizing the energy attenuation using the rheological compensation coefficient and performing time-domain correction of the transmission time delay using the reference grease injection flow rate includes:

[0043] The calculated energy attenuation is normalized using the rheological compensation coefficient.

[0044] The theoretical transmission time of the acoustic signal is calculated based on the reference grease injection flow rate, and the transmission time delay is corrected according to the difference between the theoretical transmission time and the actual calculated transmission time delay.

[0045] Preferably, the step of comparing the feature matrix with bubble feature patterns in a preset pattern library for similarity, and identifying it as a bubble event and estimating its volume if the similarity is greater than a preset threshold, includes:

[0046] The cosine similarity algorithm is used to calculate the similarity between the feature matrix and each bubble feature pattern in the preset pattern library;

[0047] If the similarity is greater than a preset threshold, it is identified as a bubble event, and the estimated volume of the bubble event is determined based on the energy decay characteristics corresponding to the matched bubble feature pattern and through a preset volume-energy decay mapping relationship.

[0048] Preferably, the step of real-time correction of the baseline grease injection flow rate, subtracting the flow rate corresponding to the estimated volume of the bubble event, to obtain the real-time grease injection flow rate includes:

[0049] When a bubble event is identified, the instantaneous flow rate corresponding to the bubble is calculated based on the estimated volume and the transit time of the bubble between the acoustic sensor arrays.

[0050] During the transit time of the bubble, the instantaneous flow rate corresponding to the bubble is subtracted from the baseline grease injection flow rate to obtain the real-time grease injection flow rate.

[0051] This invention integrates pressure, temperature, and acoustic array signals, considering changes in the rheological properties of grease due to variations in operating conditions and its slippage effect on the pipe wall. A baseline flow rate calculation model is established, and the spatiotemporal characteristics constructed from acoustic signals are used to identify and estimate the volume of air bubbles mixed in the pipeline. By subtracting the bubble volume from the baseline flow rate, the true grease injection flow rate, free from bubble interference, is obtained. This solves the technical problem that traditional metering methods cannot represent the influence of bubbles and easily lead to insufficient lubrication. It improves the reliability of grease injection flow rate monitoring and ensures that bearings receive sufficient lubrication. Attached Figure Description

[0052] Figure 1 A flowchart of the first embodiment;

[0053] Figure 2 This is a schematic diagram illustrating the relationship between the slip compensation coefficient and the historical factor.

[0054] Figure 3 This is a schematic diagram of a feedforward neural network structure;

[0055] Figure 4 This is a schematic diagram for calculating the baseline grease injection flow rate;

[0056] Figure 5 This is a schematic diagram illustrating real-time grease injection flow rate correction. Detailed Implementation

[0057] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0058] In the first embodiment, the present invention proposes a sensor-based method for real-time monitoring of bearing grease flow, such as... Figure 1 This includes the following steps:

[0059] S1, acquire the real-time pressure sequence of the pressure sensor installed in the grease injection line, the real-time temperature sequence of the temperature sensor, and the acoustic signal sequence of the acoustic sensor array arranged along the axial direction of the grease injection line.

[0060] Specifically, a piezoresistive pressure sensor and a sheathed thermocouple temperature sensor are installed on the pipeline near the inlet and outlet of the grease pump. A piezoelectric acoustic sensor is evenly distributed axially every ten centimeters along the direction of the grease injection pipeline, resulting in an array containing at least three sensors. The voltage signals from each sensor are synchronously acquired at a sampling rate of one megahertz using a data acquisition card, and these signals are converted into corresponding pressure value sequences, temperature value sequences, and multi-channel acoustic signal time-domain waveform sequences.

[0061] S2, based on the grease injection interval duration and the estimated volume of bubbles identified in the previous cycle, calculate the wall adhesion history factor, and determine the slip compensation coefficient based on the wall adhesion history factor; based on the pressure change rate of the real-time pressure sequence, the real-time temperature value, and the wall adhesion history factor, calculate the rheological compensation coefficient through a preset multivariate nonlinear mapping model; combine the real-time pressure sequence, the rheological compensation coefficient, and the slip compensation coefficient to calculate the baseline grease injection flow rate;

[0062] Record the time difference between the end of this liposuction and the end of the previous liposuction as the liposuction interval. Total volume of air bubbles identified in the previous grease injection cycle The wall adhesion history factor is a dimensionless parameter representing the state of the residual grease layer on the pipe wall, calculated using the following formula: The slip compensation coefficient is calculated by inversely from the wall adhesion history factor based on the linear relationship calibrated in the experiment. For example, the slip compensation coefficient is equal to the product of a constant - the history factor and another constant.

[0063] The preset model is a pre-trained three-layer feedforward neural network. The pressure change rate calculated from the real-time pressure sequence through differential calculation, the real-time acquired temperature value, and the calculated wall adhesion history factor are used as the three input neurons of this neural network. After processing through hidden layers, the network outputs a single value, the rheological compensation coefficient, representing the influence of grease viscosity and shear thinning characteristics under the current operating conditions. Based on the pressure values ​​in the real-time pressure sequence, a theoretical flow rate is calculated using the classic Hagen-Poiseuille law. This theoretical flow rate is multiplied by the rheological compensation coefficient to correct for errors caused by the non-Newtonian fluid characteristics of the grease. The obtained flow rate value is added to a gain term determined by the slip compensation coefficient, which is proportional to the slip compensation coefficient and the pipe cross-sectional area, to compensate for velocity slippage at the pipe wall, thus obtaining a pure grease reference flow rate that excludes the influence of air bubbles.

[0064] In an optional embodiment, the step of calculating the wall adhesion history factor based on the grease injection interval duration and the estimated volume of bubbles identified in the previous cycle, and determining the slip compensation coefficient based on the wall adhesion history factor, includes:

[0065] The wall adhesion history factor is obtained by weighting the grease injection interval duration and the estimated volume of the bubbles identified in the previous cycle.

[0066] Based on the wall adhesion history factor, the slip compensation coefficient is determined through a preset linear relationship.

[0067] The wall adhesion history factor H was obtained through weighted calculation. This calculation combined two key factors: the interval since the last grease injection. And the total volume of air bubbles identified in the previous grease injection cycle For example, if we set a weight for the duration of the interval. The weight of the bubble volume in the previous period is 0.001. The value is set to 0.8, making H a dimensionless value. The current monitored interval between injections is 300 seconds, and the volume of air bubbles identified in the previous cycle was 0.5 mL. Therefore, the wall adhesion history factor H is calculated using the following formula. Substituting the data yields H=0.7. This factor represents the change in the state of the grease layer on the pipe wall that may result from prolonged settling and historical air bubble residue. Those skilled in the art should understand that... and Although the dimensions are different, by adjusting... and This makes their contributions to calculating the wall adhesion history factor H comparable. Alternatively, one could first calculate the wall adhesion history factor H. and Perform a normalization operation, and then calculate the weighted sum.

[0068] Based on the calculated wall adhesion history factor H, the slip compensation coefficient is determined through a pre-defined linear relationship. The linear relationship is pre-calibrated in experiments or simulations, and takes the form of... Where a and b are preset constants. For example, preset parameters a is 0.5 and b is 0.75. Then the slip compensation coefficient... The value is 1.1. This coefficient is used to compensate for flow estimation errors caused by wall slippage in subsequent flow calculations. The larger the historical factor, the more pronounced the slippage effect, and the corresponding compensation coefficient value is adjusted. Figure 2 .

[0069] In an optional embodiment, the calculation of the rheological compensation coefficient based on the pressure change rate of the real-time pressure sequence, the real-time temperature value, and the wall adhesion history factor using a preset multivariate nonlinear mapping model includes:

[0070] The multivariate nonlinear mapping model is a feedforward neural network. The inputs of the network are the pressure change rate of the real-time pressure sequence, the real-time temperature value, and the wall adhesion history factor, and the output is the rheological compensation coefficient.

[0071] Feedforward neural network models are used to process and map complex nonlinear relationships. The network structure of the model described above consists of an input layer, two hidden layers, and an output layer, as shown below. Figure 3 The input layer has three neurons, each receiving one of three input variables: the real-time pressure change rate, the real-time temperature value, and the wall adhesion history factor. For example, the input data at a given moment could be a pressure change rate of 1.5 kPa / s, a temperature of 45°C, and the wall adhesion history factor of 0.7 calculated in the previous steps. Before being input into the network, the data is normalized and scaled to the range of 0 to 1.

[0072] After data input, during forward propagation, the data flows through two hidden layers. The first hidden layer has 16 neurons, and the second hidden layer has 8 neurons. Both hidden layers use the Modified Linear Unit (ReLU) as the activation function to enhance the network's nonlinear expressive power. Each neuron in each layer performs a nonlinear transformation on the weighted sum of the outputs of the previous layer. The signal is then transmitted to the output layer, which contains only one neuron and uses a linear activation function to directly output a scalar value, which is the rheological compensation coefficient. For example, given the input above, the network might output: The value is 1.2, and the coefficient reflects the change in the rheological properties of the grease under the current operating conditions.

[0073] The feedforward neural network takes as input a three-dimensional feature vector composed of real-time pressure change rate, real-time temperature value, and wall adhesion history factor, and outputs a single scalar value, namely the rheological compensation coefficient. The network structure of the feedforward neural network is a fully connected network containing one input layer, two hidden layers, and one output layer. The training set data is obtained through numerous lubricating grease flow physics experiments or high-fidelity computational fluid dynamics simulations under different operating conditions. Each data sample contains a specific set of operating condition parameters and corresponding labels of the true rheological compensation coefficients obtained through precise measurement or calculation. The training process uses the Adam optimizer, adjusting the network weights through the backpropagation algorithm. The goal is to minimize the mean squared error loss function L between the network's predicted values ​​and the true labels, defined as follows:

[0074]

[0075] Where N is the batch size. It is the true value of the i-th sample, and The "hat" represents the network's predicted value. Training will continue for multiple epochs until the model's performance on independent validation datasets no longer improves, marking the completion of model training.

[0076] In an optional embodiment, calculating the baseline grease injection flow rate by combining the real-time pressure sequence, the rheological compensation coefficient, and the slip compensation coefficient includes:

[0077] The pressure difference ΔP between the inlet and outlet of the grease injection line is calculated based on the real-time pressure sequence, and combined with the line diameter D, line length L, grease base viscosity μ, and the rheological compensation coefficient. and the slip compensation coefficient The baseline grease injection flow rate is calculated according to the following formula. : .

[0078] Real-time pressure data is acquired from pressure sensors deployed at both ends of the grease injection line to calculate the pressure difference ΔP between the inlet and outlet of the line; for example, ΔP is 500 kPa. The geometric parameters of the line and the basic physical properties of the grease are obtained, typically known or preset, such as a line diameter D of 0.01 m, a line length L of 2 m, and a basic viscosity μ of 10 Pa·s. The compensation coefficient, i.e., the slip compensation coefficient, calculated by other modules, is also obtained. The rheological compensation coefficient is 1.1. The value is 1.2. After obtaining all parameters, the values ​​are substituted into the modified Hagen-Poiseuille equation for calculation. This equation is based on the classical Hagen-Poiseuille equation, with two compensation coefficients added. and It adapts to the flow characteristics of non-Newtonian fluids under complex working conditions. The baseline grease injection flow rate was calculated. Approximately 5.625 mL / s. The obtained flow rate value is an estimate of the theoretical grease injection flow rate under current operating conditions with no air bubbles present, such as... Figure 4 .

[0079] S3, calculate the transmission time delay and energy attenuation of the acoustic signal between adjacent sensors; normalize the energy attenuation using the rheological compensation coefficient, and perform time-domain correction on the transmission time delay using the reference grease injection flow rate; construct a feature matrix based on the normalized energy attenuation and the corrected transmission time delay; compare the similarity of the feature matrix with the bubble feature patterns in the preset pattern library, and if the similarity is greater than a preset threshold, identify it as a bubble event and estimate its volume;

[0080] Specifically, cross-correlation is performed on the signals from two adjacent sensors in the acoustic array, and the time point corresponding to the peak of the cross-correlation function is the transmission time delay. The root mean square values ​​of the two signals are calculated separately, and their ratio is the energy attenuation. The energy attenuation value is divided by a function value positively correlated with the rheological compensation coefficient to complete normalization. The transmission time delay is subtracted from the time obtained by dividing the sensor spacing by the quotient of the reference grease injection flow rate and the pipe cross-sectional area to complete the time-domain correction. The normalized energy attenuation and corrected transmission time delay of all adjacent sensor pairs are combined into a feature vector as the feature matrix. The preset pattern library stores the standard feature matrices generated by bubbles of different volumes under standard operating conditions. By calculating the cosine similarity between the current feature matrix and all patterns in the library, if the maximum similarity exceeds 0.9, it is determined to be a bubble event, and the bubble volume in the pattern library corresponding to the maximum similarity is used as the estimated volume of the current bubble.

[0081] In an optional embodiment, the calculation of the acoustic signal transmission time delay and energy attenuation between adjacent sensors includes:

[0082] The transmission time delay between adjacent sensor signals in the acoustic signal sequence is calculated using the generalized cross-correlation-phase transform method.

[0083] The energy attenuation of the signal is calculated based on the root mean square energy of the acoustic signals from the adjacent sensors.

[0084] To calculate the transmission time delay, synchronized acoustic signal sequences are acquired from two adjacent acoustic sensors. and The Generalized Cross-Correlation-Phase Transform (GCC-PHAT) algorithm is employed. This algorithm calculates the cross-power spectrum of the two signals and uses phase information for spectral whitening, which can suppress noise interference and improve the accuracy of time delay estimation. The calculation process includes performing a Fourier transform on the signals, calculating the cross-power spectrum, performing a phase transform, and obtaining the cross-correlation function through an inverse Fourier transform. The time corresponding to the peak position of this function is the transmission time delay τ of the signal between the two sensors. For example, if the peak of the cross-correlation function occurs at the 285th sampling point with a sampling frequency of 100 Hz, the transmission time delay is 2.85 s, which represents the time required for the acoustic characteristics of the bubble to propagate from the upstream sensor to the downstream sensor.

[0085] To calculate energy decay, the signal sequence within the same time window is analyzed. and Calculate the root mean square energy respectively. and The signal energy of the upstream sensor is calculated. The signal energy of the downstream sensor is 0.5V. The value is 0.4V. The energy decay A is calculated based on these two energy values, typically expressed in decibels (dB), using the following formula:

[0086]

[0087] Substituting the data, we get that the energy attenuation A ≈ 1.94 dB. This value reflects the energy loss of the acoustic signal during propagation due to absorption and scattering by the medium. The presence of bubbles will affect this attenuation value.

[0088] In an optional embodiment, normalizing the energy attenuation using the rheological compensation coefficient and performing time-domain correction of the transmission time delay using the reference grease injection flow rate includes:

[0089] The calculated energy attenuation is normalized using the rheological compensation coefficient.

[0090] The theoretical transmission time of the acoustic signal is calculated based on the reference grease injection flow rate, and the transmission time delay is corrected according to the difference between the theoretical transmission time and the actual calculated transmission time delay.

[0091] Changes in the rheological state of the grease itself can affect the attenuation of acoustic signals. To eliminate this interference and highlight the abnormal attenuation caused by air bubbles, a rheological compensation coefficient output by a feedforward neural network is used. For example, in step 1.2, the original energy decay value A, for example, 1.94 dB, is normalized. The normalization operation is to divide A by... Normalized energy decay is obtained The calculated result is approximately 1.62. This normalized value... It can more purely reflect the signal attenuation caused by abnormal events in the pipeline, such as air bubbles.

[0092] Based on baseline fat injection flow rate Calculate the theoretical transport time of the bubble Given that the sensor spacing d is 0.2m and the pipe diameter D is 0.01m, the pipe cross-sectional area S can be calculated. The average flow velocity v = / S, if If the flow rate is 5.625 × 10⁻⁶ m³ / s, then the flow velocity v is approximately 0.0716 m / s. Theoretical transport time. =d / v, approximately 2.79s. This theoretical value is compared to the actual measured transmission time delay τ, for example, 2.85s. The difference Δτ is 0.06s, reflecting the deviation between the actual flow velocity and the theoretical average flow velocity. This deviation information is used to generate a corrected transmission time delay. For example, the measured value of 2.85s can be directly used, while the deviation of 0.06s is recorded for subsequent feature analysis and model adjustment.

[0093] In an optional embodiment, the step of comparing the feature matrix with bubble feature patterns in a preset pattern library for similarity, and identifying it as a bubble event and estimating its volume if the similarity is greater than a preset threshold, includes:

[0094] The cosine similarity algorithm is used to calculate the similarity between the feature matrix and each bubble feature pattern in the preset pattern library;

[0095] If the similarity is greater than a preset threshold, it is identified as a bubble event, and the estimated volume of the bubble event is determined based on the energy decay characteristics corresponding to the matched bubble feature pattern and through a preset volume-energy decay mapping relationship.

[0096] Corrected transmission time delay from all sensor pairs and normalized energy decay These are combined into a spatiotemporal feature vector. For example, for a system containing two sensor pairs, the feature vector at the current moment... Possible values ​​are [2.85, 1.62, 2.88, 1.59]. Maintain a pre-built pattern library that stores standard feature vectors corresponding to bubbles of different sizes and shapes; for example, the standard feature vector of a 2mL bubble. The values ​​are [2.80, 1.65, 2.82, 1.63]. A cosine similarity algorithm is used to calculate... With each in the library The similarity score is calculated, for example, 0.98, and compared with a preset threshold, such as 0.95. Since 0.98 > 0.95, the currently detected event is determined to be highly matched with the 2mL bubble pattern, and therefore identified as a bubble event.

[0097] After successfully identifying bubble events and matching the most similar feature pattern, the volume estimation stage begins. This stage utilizes a pre-calibrated volume-energy decay mapping relationship, which can be a lookup table or a mathematical function, representing the relationship between the bubble volume V and the normalized energy decay. The quantitative relationship between them.

[0098] Extract the current feature vector The average normalized energy decay value is 1.605. This value is then substituted into a preset mapping function, such as a quadratic function model, to calculate the estimated volume of the bubble event. If the function calculation result is 3.64, the estimated volume of the bubble is determined to be 3.64 mL.

[0099] S4, the baseline grease injection flow rate is corrected in real time, and the flow rate corresponding to the estimated volume of the bubble event is deducted to obtain the real-time grease injection flow rate.

[0100] Specifically, once a bubble event is identified, the estimated bubble volume is divided by the duration of the event on the acoustic signal to obtain the equivalent volumetric flow rate of the bubble as it flows through the pipeline. During the duration of the bubble event, the equivalent volumetric flow rate is subtracted from the calculated baseline grease injection flow rate. Outside of this period, the real-time grease injection flow rate equals the baseline grease injection flow rate. This process is repeated to continuously correct the flow rate throughout the entire grease injection process.

[0101] In an optional embodiment, the real-time correction of the baseline grease injection flow rate by subtracting the flow rate corresponding to the estimated volume of the bubble event to obtain the real-time grease injection flow rate includes:

[0102] When a bubble event is identified, the instantaneous flow rate corresponding to the bubble is calculated based on the estimated volume and the transit time of the bubble between the acoustic sensor arrays.

[0103] During the transit time of the bubble, the instantaneous flow rate corresponding to the bubble is subtracted from the baseline grease injection flow rate to obtain the real-time grease injection flow rate.

[0104] Obtain the estimated volume of the bubble. For example, 3.64 mL, which is 3.64 × 10⁻⁶ m³. Calculate the total transit time required for the bubble to pass through the entire sensor array. The transit time is obtained by summing the corrected transmission time delays between all adjacent sensor pairs. For example, if there are two sensor pairs with time delays of 2.85s and 2.88s respectively, the total transit time is 5.73s. The instantaneous flow rate represented by the bubble itself is then calculated. ,Right now Divide by Substituting the data, we can calculate... It is approximately 0.635 × 10⁻⁶ m³ / s.

[0105] After obtaining the instantaneous flow rate of the bubbles, the baseline grease injection flow rate is used. Real-time corrections are performed. Baseline grease injection flow rate. This is the previously calculated theoretical grease flow rate, for example, 5.625 × 10⁻⁶ m³ / s. The real-time grease injection flow rate is measured within the 5.73 s time window during which air bubbles pass through the sensor array. Calculated by subtracting the bubble flow rate from the baseline flow rate. .

[0106] The calculated result is 4.99 × 10⁻⁶ m³ / s. This corrected flow rate value better reflects the actual volumetric flow rate of the grease delivered during this time period because it excludes the contribution of gas volume. Once the air bubbles have completely passed through the monitoring area, i.e., outside the transit time window, the real-time grease injection flow rate reverts to the baseline grease injection flow rate. ,like Figure 5 .

[0107] In a second embodiment, the present invention also provides a sensor-based real-time monitoring system for bearing grease injection flow, comprising the following modules:

[0108] The acquisition module is used to acquire the real-time pressure sequence of the pressure sensor installed in the grease injection line, the real-time temperature sequence of the temperature sensor, and the acoustic signal sequence of the acoustic sensor array arranged along the axial direction of the grease injection line.

[0109] The calculation module is used to calculate the wall adhesion history factor based on the grease injection interval duration and the estimated volume of bubbles identified in the previous cycle, and to determine the slip compensation coefficient based on the wall adhesion history factor; to calculate the rheological compensation coefficient based on the pressure change rate of the real-time pressure sequence, the real-time temperature value, and the wall adhesion history factor through a preset multivariate nonlinear mapping model; and to calculate the baseline grease injection flow rate by combining the real-time pressure sequence, the rheological compensation coefficient, and the slip compensation coefficient.

[0110] The identification module is used to calculate the transmission time delay and energy attenuation of acoustic signals between adjacent sensors; normalize the energy attenuation using the rheological compensation coefficient, and perform time-domain correction on the transmission time delay using the reference grease injection flow rate; construct a feature matrix based on the normalized energy attenuation and the corrected transmission time delay; compare the similarity of the feature matrix with the bubble feature patterns in the preset pattern library, and if the similarity is greater than a preset threshold, it is identified as a bubble event and the volume is estimated.

[0111] The correction module is used to correct the baseline grease injection flow rate in real time, deduct the flow rate corresponding to the estimated volume of the bubble event, and obtain the real-time grease injection flow rate.

[0112] In this specification, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise limited, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element. In this document, "a," "an," "the," "the," and "its" may also include plural forms unless the context clearly indicates otherwise. "Multiple" refers to at least two, such as 2, 3, 5, or 8, etc. "And / or" includes any and all combinations of the associated listed items.

[0113] The various embodiments in this specification are described in a progressive manner. Each embodiment focuses on the differences from other embodiments. The various embodiments can be combined as needed, and the same or similar parts can be referred to each other.

[0114] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A sensor-based method for real-time monitoring of bearing grease flow rate, characterized in that, Includes the following steps: The real-time pressure sequence of the pressure sensor installed in the grease injection line, the real-time temperature sequence of the temperature sensor, and the acoustic signal sequence of the acoustic sensor array arranged along the axial direction of the grease injection line are obtained. Based on the grease injection interval duration and the estimated volume of bubbles identified in the previous cycle, the wall adhesion history factor is calculated, and the slip compensation coefficient is determined based on the wall adhesion history factor; based on the pressure change rate of the real-time pressure sequence, the real-time temperature value, and the wall adhesion history factor, the rheological compensation coefficient is calculated through a preset multivariate nonlinear mapping model; combined with the real-time pressure sequence, the rheological compensation coefficient, and the slip compensation coefficient, the baseline grease injection flow rate is calculated; Calculate the transmission time delay and energy attenuation of acoustic signals between adjacent sensors; The energy attenuation is normalized using the rheological compensation coefficient, and the transmission time delay is corrected in the time domain using the reference grease injection flow rate. A feature matrix is ​​constructed based on the normalized energy attenuation and the corrected transmission time delay. The feature matrix is ​​compared with the bubble feature patterns in the preset pattern library. If the similarity is greater than a preset threshold, it is identified as a bubble event and the volume is estimated. The baseline grease injection flow rate is corrected in real time, and the flow rate corresponding to the estimated volume of the bubble event is deducted to obtain the real-time grease injection flow rate.

2. The method according to claim 1, characterized in that, The calculation of the wall adhesion history factor based on the grease injection interval duration and the estimated volume of bubbles identified in the previous cycle, and the determination of the slip compensation coefficient based on the wall adhesion history factor, includes: The wall adhesion history factor is obtained by weighting the grease injection interval duration and the estimated volume of the bubbles identified in the previous cycle. Based on the wall adhesion history factor, the slip compensation coefficient is determined through a preset linear relationship.

3. The method according to claim 1, characterized in that, The calculation of rheological compensation coefficients based on the pressure change rate of the real-time pressure sequence, the real-time temperature value, and the wall adhesion history factor using a preset multivariate nonlinear mapping model includes: The multivariate nonlinear mapping model is a feedforward neural network. The inputs of the network are the pressure change rate of the real-time pressure sequence, the real-time temperature value, and the wall adhesion history factor, and the output is the rheological compensation coefficient.

4. The method according to claim 3, characterized in that, The calculation of the baseline grease injection flow rate by combining the real-time pressure sequence, the rheological compensation coefficient, and the slip compensation coefficient includes: The pressure difference ΔP between the inlet and outlet of the grease injection line is calculated based on the real-time pressure sequence, and combined with the line diameter D, line length L, grease base viscosity μ, and the rheological compensation coefficient. and the slip compensation coefficient The baseline grease injection flow rate is calculated according to the following formula. : .

5. The method according to claim 1, characterized in that, The calculation of the acoustic signal transmission time delay and energy attenuation between adjacent sensors includes: The transmission time delay between adjacent sensor signals in the acoustic signal sequence is calculated using the generalized cross-correlation-phase transform method. The energy attenuation of the signal is calculated based on the root mean square energy of the acoustic signals from the adjacent sensors.

6. The method according to claim 1, characterized in that, The step of normalizing the energy attenuation using the rheological compensation coefficient and performing time-domain correction of the transmission time delay using the reference grease injection flow rate includes: The calculated energy attenuation is normalized using the rheological compensation coefficient. The theoretical transmission time of the acoustic signal is calculated based on the reference grease injection flow rate, and the transmission time delay is corrected according to the difference between the theoretical transmission time and the actual calculated transmission time delay.

7. The method according to claim 1, characterized in that, The step of comparing the feature matrix with bubble feature patterns in a preset pattern library for similarity, and identifying it as a bubble event and estimating its volume if the similarity is greater than a preset threshold, includes: The cosine similarity algorithm is used to calculate the similarity between the feature matrix and each bubble feature pattern in the preset pattern library; If the similarity is greater than a preset threshold, it is identified as a bubble event, and the estimated volume of the bubble event is determined based on the energy decay characteristics corresponding to the matched bubble feature pattern and through a preset volume-energy decay mapping relationship.

8. The method according to claim 1, characterized in that, The real-time correction of the baseline grease injection flow rate, by subtracting the flow rate corresponding to the estimated volume of the bubble event, to obtain the real-time grease injection flow rate includes: When a bubble event is identified, the instantaneous flow rate corresponding to the bubble is calculated based on the estimated volume and the transit time of the bubble between the acoustic sensor arrays. During the transit time of the bubble, the instantaneous flow rate corresponding to the bubble is subtracted from the baseline grease injection flow rate to obtain the real-time grease injection flow rate.

9. A sensor-based real-time monitoring system for bearing grease injection flow, characterized in that, Includes the following modules: The acquisition module is used to acquire the real-time pressure sequence of the pressure sensor installed in the grease injection line, the real-time temperature sequence of the temperature sensor, and the acoustic signal sequence of the acoustic sensor array arranged along the axial direction of the grease injection line. The calculation module is used to calculate the wall adhesion history factor based on the grease injection interval duration and the estimated volume of the bubbles identified in the previous cycle, and to determine the slip compensation coefficient based on the wall adhesion history factor. Based on the pressure change rate of the real-time pressure sequence, the real-time temperature value, and the wall adhesion history factor, the rheological compensation coefficient is calculated through a preset multivariate nonlinear mapping model. The baseline grease injection flow rate is calculated by combining the real-time pressure sequence, the rheological compensation coefficient, and the slip compensation coefficient. The identification module is used to calculate the transmission time delay and energy attenuation of acoustic signals between adjacent sensors; The energy attenuation is normalized using the rheological compensation coefficient, and the transmission time delay is corrected in the time domain using the reference grease injection flow rate. A feature matrix is ​​constructed based on the normalized energy attenuation and the corrected transmission time delay. The feature matrix is ​​compared with the bubble feature patterns in the preset pattern library. If the similarity is greater than a preset threshold, it is identified as a bubble event and the volume is estimated. The correction module is used to correct the baseline grease injection flow rate in real time, deduct the flow rate corresponding to the estimated volume of the bubble event, and obtain the real-time grease injection flow rate.

10. The system according to claim 9, characterized in that, The calculation of the wall adhesion history factor based on the grease injection interval duration and the estimated volume of bubbles identified in the previous cycle, and the determination of the slip compensation coefficient based on the wall adhesion history factor, includes: The wall adhesion history factor is obtained by weighting the grease injection interval duration and the estimated volume of the bubbles identified in the previous cycle. Based on the wall adhesion history factor, the slip compensation coefficient is determined through a preset linear relationship.

Images