A cam follower bearing wear extent monitoring method and system

Abstract

The present application belongs to the technical field of bearing vibration analysis test, and particularly relates to a cam follower bearing wear degree monitoring method and system, which comprises the following steps: using a rotating speed signal to perform equal-angle resampling processing on a vibration signal, using a variational mode decomposition algorithm to decompose the obtained angular domain vibration signal into a plurality of modal components, evaluating the impact significance weight of each modal component according to the kurtosis value of each modal component and the correlation coefficient of each modal component and the original vibration signal, and obtaining a comprehensive impact energy by energy-weighted summation of each modal component; using the comprehensive impact energy and the vibration energy calculated by a physical mapping model to calculate a speed decoupling wear index that removes the rotating speed influence; and comparing the speed decoupling wear index with a preset adaptive alarm threshold, and determining the wear state of the cam follower bearing according to the comparison result. The present application solves the false alarm and missed alarm problems under variable speed conditions.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of vibration analysis test of bearings. More particularly, the present application relates to a cam follower bearing wear degree monitoring method and system. BACKGROUND

[0002] As a core transmission component in automated production lines, industrial robot joints and cam indexing mechanisms, the cam follower bearing has the structural characteristics of a thick outer ring wall, which directly rolls on the cam surface or guide rail as a roller, and a precise needle inside to bear high radial load. In actual industrial applications, the working conditions of such bearings are often highly non-stationary, that is, their rotational speed and load will change dramatically and periodically with the production rhythm, such as frequent start-up acceleration, high-speed operation, braking deceleration, etc. This complex operating environment poses great challenges to bearing condition monitoring.

[0003] The existing bearing wear monitoring technology usually uses high-frequency piezoelectric vibration sensors to collect vibration signals and calculates the effective value, peak value or energy spectrum density of the full frequency band, and the monitoring logic usually sets a fixed empirical threshold. When the monitored vibration energy index exceeds the threshold, the system determines that the bearing is worn and triggers an alarm.

[0004] However, this fixed threshold-based full-band energy analysis method has significant limitations under non-stationary conditions: on the one hand, according to the principles of mechanical dynamics, the vibration energy of the bearing is positively correlated with the rotational speed. During high-speed operation of the equipment, even if the bearing is completely healthy, the vibration energy amplitude it produces can be very high, easily breaking through the pre-set fixed alarm threshold, causing the monitoring system to misjudge the normal high-speed operation as bearing wear, resulting in false high-energy false alarms; on the other hand, the early wear characteristics of the cam follower bearing internal needle, such as pitting and micro-peeling, produce extremely weak impact signals. In the presence of strong background noise, such as motor drive noise and guide rail friction noise, the traditional full-band energy analysis method cannot extract specific wear characteristics from the complex mixed signals, and the early fault signals are easily overwhelmed by background noise, making the system unable to identify the abnormal state of the bearing in time, resulting in feature-overshadowing false negatives. SUMMARY

[0005] To solve the technical problems of false positives and false negatives of the above-mentioned prior art under variable speed non-stationary conditions, the present application provides solutions in the following aspects.

[0006] In a first aspect, the present invention provides a method for monitoring the wear degree of a cam follower bearing, comprising: simultaneously acquiring the original vibration signal and rotational speed signal of the cam follower bearing, and using the rotational speed signal to perform equal-angle resampling processing on the vibration signal to obtain an angular domain vibration signal; decomposing the angular domain vibration signal into several modal components using a variational mode decomposition algorithm, evaluating the impact significance weight of each modal component based on the kurtosis value of each modal component and the correlation coefficient between each modal component and the original vibration signal, and weighting and summing the energy of each modal component based on the impact significance weight to obtain a comprehensive impact energy; constructing a physical mapping model for calculating vibration energy using the rotational speed signal based on a preset health benchmark coefficient; and calculating a velocity decoupling wear index that removes the influence of rotational speed using the comprehensive impact energy and the vibration energy calculated by the physical mapping model. ,and , Indicates the total impact energy. Indicates the rotational speed signal. , The health baseline coefficient is represented; the speed decoupling wear index is compared with the preset adaptive alarm threshold, and the wear state of the cam follower bearing is determined based on the comparison result.

[0007] This invention constructs a speed-decoupled wear index and uses a physical model to accurately fit the law of vibration energy variation with rotational speed. When the equipment is running at high speed, the index can automatically deduct the energy gain caused by speed, keeping the healthy bearing index at a low level. When running at low speed, the wear increment is highlighted through normalization, which makes the monitoring index stable across the entire speed range and no longer fluctuate with rotational speed, solving the problem of false alarms and missed alarms under variable speed conditions. In addition, this invention no longer blindly calculates the energy of all frequency bands, but focuses on those frequency bands with high kurtosis and high correlation through the calculation of impact significance weight. In the early stage of needle roller wear, this weighting mechanism can amplify weak impact signals, significantly improve the signal-to-noise ratio of weak wear characteristics, make them stand out in noisy industrial background noise, and greatly advance the time window for fault detection.

[0008] Preferably, the formula for calculating the impact significance weight of the modal component is as follows: In the formula: Indicates the first Impact significance weights for each modal component; , They represent the first The and the first One modal component; , They represent the first The and the first kurtosis values ​​of each modal component; Represents the original vibration signal; The number of all modal components obtained by the decomposition is equal to ; , The Pearson correlation coefficient of the first modal component and the original vibration signal is represented as The Pearson correlation coefficient of the first modal component and the original vibration signal is represented as The absolute value is represented as

[0009] The present application can evaluate the probability of containing effective fault impact components in each modal component by introducing the kurtosis value representing the impact and the Pearson correlation coefficient representing the correlation with the original signal, which enables the algorithm to effectively distinguish the periodic impact signal caused by wear from the random outlier noise with simply high kurtosis, thereby automatically suppressing the interference frequency band during subsequent synthesis energy, and ensuring that the monitoring index is highly focused on the real wear characteristics of the bearing.

[0010] Preferably, the calculation formula of the comprehensive impact energy is: ; in the formula: The comprehensive impact energy is represented as The number of all modal components obtained by the decomposition is equal to ; The impact significance weight of the first modal component is represented as The length of the angular domain vibration signal is represented as, and all modal components are equal to ; The first discrete signal point of the first modal component is represented as .

[0011] The present application uses the impact significance weight to perform weighted summation on the energy of each component after the variational modal decomposition, and this weighting mechanism can amplify the frequency band energy weight containing fault information, further highlight the weak impact energy generated by early wear, provide a high signal-to-noise ratio data basis for subsequent decoupling calculation, and improve the sensitivity to early faults.

[0012] Preferably, the acquisition method of the health reference coefficient is: using the historical running data of the equipment in the healthy state, fitting the reference curve of the vibration energy of the bearing with the change of the rotating speed by the least square method, and the reference curve belongs to a one-dimensional quadratic curve not containing a linear term, only the quadratic term and the constant term are contained in the fitted reference curve, the coefficient of the quadratic term obtained is taken as the health reference coefficient , and the constant term obtained is taken as the health reference coefficient . characterizing the proportionality coefficient of the vibration energy growing with the square of the rotational speed, characterizing the inherent floor noise energy of the system at standstill or very low speed.

[0013] Preferably, the original vibration signal is collected by a high-frequency piezoelectric vibration sensor, and the rotational speed signal is synchronously collected by an industrial field bus or an encoder interface of a servo driver.

[0014] Preferably, the equal-angle resampling of the vibration signal using the rotational speed signal to obtain an angular domain vibration signal comprises: synchronously collecting vibration time series and corresponding rotational speed time series within a period of time; performing a detrending process on the vibration time series to eliminate the direct current component and baseline drift; performing equal-angle resampling on the vibration time series using the real-time rotational speed time series to convert the time-domain non-stationary signal into an angular domain stationary signal to obtain the angular domain vibration signal.

[0015] The present application adds a detrending process before resampling to eliminate the influence of the direct current component and baseline drift; through equal-angle resampling using real-time rotational speed, the time-domain non-stationary signal is converted into an angular domain stationary signal, effectively overcoming the frequency spectrum ambiguity and amplitude modulation phenomenon caused by rotational speed fluctuation under non-stationary working conditions, so that the subsequent signal analysis is no longer limited by the change of rotational speed.

[0016] Preferably, the determination of the wear state of the cam follower bearing according to the comparison result comprises: real-time calculation of a speed decoupling wear index and comparison with an adaptive alarm threshold value : if the speed decoupling wear index is less than or equal to the adaptive alarm threshold value, it is determined that the cam follower bearing is in a healthy state; if the speed decoupling wear index is greater than the adaptive alarm threshold value, it is determined that the cam follower bearing has wear abnormality, and an alarm signal is triggered.

[0017] Preferably, the setting method of the adaptive alarm threshold value is: extracting historical records of the last one to three months and eliminating non-business noise points, calculating the speed decoupling wear index of each reference data obtained; drawing a histogram of the speed decoupling wear index of all reference data to determine its distribution form: if the histogram presents an approximate normal symmetric distribution, the absolute median difference method is used to determine the candidate threshold value, and if the histogram presents a significant long tail or skew distribution, the quantile method is used to determine the candidate threshold value; the obtained candidate threshold value is substituted into the reference data stream for simulation running, the false alarm rate in the known normal period and the sensitivity in the fault period are counted, and according to the false alarm rate and the sensitivity, the coefficients in the absolute median difference method or the quantile method are dynamically adjusted, so that the obtained candidate threshold value is adjusted until the best balance point between the false alarm rate and the sensitivity is found, and the corresponding candidate threshold value is taken as the adaptive alarm threshold value.

[0018] In a second aspect, the present application provides a cam follower bearing wear degree monitoring system, comprising a processor and a memory, the memory storing computer program instructions which, when executed by the processor, implement the above-mentioned cam follower bearing wear degree monitoring method.

[0019] By adopting the above technical solution, the above-mentioned cam follower bearing wear degree monitoring method is generated into a computer program and stored in the memory to be loaded and executed by the processor, so that a terminal device is made according to the memory and the processor, and use is facilitated.

[0020] The present application has the following beneficial effects:

[0021] The present application accurately fits the law of vibration energy changing with speed by constructing a speed-decoupled wear index: in high-speed operation of the equipment, the index can automatically deduct the energy gain caused by speed, so that the healthy bearing index is maintained at a low level; in low-speed operation, the wear increment is highlighted through normalization, which makes the monitoring index remain stable in the full speed range and no longer fluctuate with speed, solving the false alarm and missed alarm problems in variable speed working conditions; in addition, the present application no longer blindly calculates the energy of all frequency bands, but focuses on those frequency bands with high kurtosis and high correlation through calculation of impact significance weight, amplifies the weak impact signal in the early stage of needle wear, significantly improves the signal-to-noise ratio of weak wear characteristics, and makes it stand out in the noisy industrial background noise, greatly advancing the time window of fault discovery. BRIEF DESCRIPTION OF DRAWINGS

[0022] Figure 1 is a flow chart schematically showing a cam follower bearing wear degree monitoring method in the present application;

[0023] Figure 2 is a schematic diagram schematically showing real-time speed signal and original vibration signal in a working cycle;

[0024] Figure 3 is a schematic diagram schematically showing bearing monitoring and early warning by the prior art through a traditional energy monitoring index;

[0025] Figure 4 is a schematic diagram schematically showing bearing monitoring and early warning by the present application through a speed-decoupled wear index. DETAILED DESCRIPTION

[0026] With reference to the drawings of the embodiments of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments of the present application, all the other embodiments obtained by a person of ordinary skill in the art without creative work are within the protection scope of the present application.

[0027] The specific embodiments of the present application will be described in detail below with reference to the drawings.

[0028] The embodiments of the present application disclose a cam follower bearing wear degree monitoring method, referring to Figure 1 , comprising steps S1-S4:

[0029] S1, synchronously collect the vibration signal and the rotating speed signal of the cam follower bearing, and utilize the rotating speed signal to perform equal-angle resampling processing on the vibration signal to obtain an angular domain vibration signal.

[0030] It should be noted that the cam follower bearing operates under non-stationary conditions, and the rotating speed and the load thereof fluctuate dramatically with production rhythm, which directly leads to frequency ambiguity and amplitude modulation phenomenon of the vibration signal. If the time domain signal is directly analyzed, it is difficult to distinguish the energy change caused by the rotating speed fluctuation and the fault impact.

[0031] Specifically, a high-frequency piezoelectric vibration sensor is installed at the fixed seat position of the cam follower bearing to collect the original vibration signal of the equipment. Meanwhile, the rotating speed signal of the equipment is synchronously collected through an industrial field bus or an encoder interface of a servo driver. The sampling frequency is set to be higher than 20 kHz to ensure covering the high-frequency impact characteristics of the cam follower bearing. The vibration time series and the corresponding rotating speed time series within a period of time are synchronously collected.

[0032] Further, the vibration time series is subjected to a de-trending item processing to eliminate the direct current component and the baseline drift. The vibration time series is subjected to equal-angle resampling by utilizing the real-time rotating speed time series to convert the time domain non-stationary signal into an angular domain stationary signal to obtain the angular domain vibration signal, so as to eliminate the frequency spectrum ambiguity caused by the rotating speed fluctuation and provide a high-quality data basis for subsequent feature extraction.

[0033] Exemplarily, Figure 2Fig. 1 is a schematic diagram of a real-time speed signal and an original vibration signal in a working cycle; wherein, the curve corresponding to the speed signal represents that the automation equipment where the cam follower bearing is located experiences a complete trapezoidal speed change process of low-speed standby (0-2 seconds), linear acceleration (2-4 seconds), high-speed processing (4-8 seconds), linear deceleration (8-10 seconds) and low-speed standby (10-12 seconds); in the curve corresponding to the original vibration signal, the amplitude of the vibration signal and the speed present a strong positive correlation, the vibration amplitude reaches the maximum in the middle high-speed stage, and the vibration amplitude is small in the low-speed stage at both ends, so that under the non-stationary working condition, the increase in energy caused by the speed rise or bearing wear cannot be distinguished only by the amplitude of the original signal.

[0034] S2, decompose the angular domain vibration signal into a plurality of modal components by using a variational mode decomposition algorithm, evaluate the impact significance weight of each modal component according to the kurtosis value of each modal component and the correlation coefficient of each modal component and the original vibration signal, and obtain the comprehensive impact energy by energy-weighted summation of each modal component based on the impact significance weight.

[0035] It should be noted that the early wear signal of the cam follower bearing is extremely weak and is often submerged in strong background noise such as motor driving noise and guide rail friction noise, and the traditional full-band energy analysis method cannot distinguish the effective impact component from the background noise, resulting in extremely low signal-to-noise ratio; therefore, a mechanism is needed to automatically filter and enhance the effective impact component from the complex mixed signal.

[0036] Specifically, the preprocessed angular domain vibration signal is decomposed into modal components by using a variational mode decomposition algorithm, denoted as to ; wherein, the modal number parameter in the variational mode decomposition algorithm has an integer value in the range of [3, 6]; the penalty factor parameter in the variational mode decomposition algorithm determines the bandwidth of the mode, and has a value in the range of [1000, 3000]; in this embodiment, the modal number parameter is 5, and the penalty factor parameter is 1800; the kurtosis value of each modal component and the correlation coefficient of each modal component and the original vibration signal are calculated respectively.

[0037] Further, based on the kurtosis value and the correlation coefficient, the impact significance weight of the th modal component is calculated, and the specific calculation formula is as follows:

[0038]

[0039] In the formula, the impact significance weight of the th modal component is denoted as , and has a value in the range of [0, 1]. , They represent the first The and the first One modal component; , They represent the first The and the first The kurtosis value of each modal component; the larger the kurtosis value, the higher the probability that the frequency band contains periodic transient impacts caused by wear. Represents the original vibration signal; The mode number parameter in the variational mode decomposition algorithm is equal to the number of all modal components obtained by decomposition. ; , Indicates the first The Pearson correlation coefficient between each modal component and the original vibration signal; Indicates the first The Pearson correlation coefficient between each modal component and the original vibration signal is used to constrain the selected features to be the main components of the original vibration signal, thus preventing the selection of random outlier noise with high kurtosis. Indicates taking the absolute value; denominator term It is the sum of the index products of all modal components, used for normalization.

[0040] Furthermore, based on the impact significance weights, the energy of each modal component is weighted and summed to obtain the comprehensive impact energy. The specific calculation formula is as follows:

[0041]

[0042] In the formula: Indicates the total impact energy; The mode number parameter in the variational mode decomposition algorithm is equal to the number of all modal components obtained by decomposition. ; Indicates the first Impact significance weights for each modal component; This represents the length of the angular domain vibration signal, and all modal components are equal to... ; Indicates the first The first modal component A discrete signal point; Represents instantaneous power.

[0043] Among them, by introducing impact significance weights The modal components of the stationary noise with low kurtosis or the interference signal with low correlation are greatly inhibited, so that the comprehensive impact energy is highly focused on the impact characteristics caused by wear; the comprehensive impact energy is calculated by using the variational mode decomposition combined with the impact significance weight, so that the weak early wear characteristics can be effectively extracted from the strong background noise, and the signal-to-noise ratio is significantly improved.

[0044] S3, based on the preset health reference coefficient, a physical mapping model for calculating vibration energy by using the rotation speed signal is constructed, and a speed decoupling wear index is calculated by using the comprehensive impact energy and the vibration energy calculated by the physical mapping model.

[0045] It should be noted that according to the principle of mechanical dynamics, the vibration energy of the bearing is positively correlated with the rotation speed, and in the high-speed operation stage of the equipment, even if the bearing is healthy, the vibration energy generated by the bearing will greatly increase, which is easy to break through the fixed alarm threshold and cause false alarm; on the contrary, in the low-speed stage, due to the low baseline energy, the weak wear characteristics are easy to be missed; therefore, a physical mapping model of energy and rotation speed must be established to strip the vibration energy naturally increased due to the increase of rotation speed, and only the vibration energy increment additionally generated due to wear defects is retained.

[0046] Specifically, according to the Hertz contact theory and rotor dynamics, the vibration energy of the bearing is approximately proportional to the square of the rotation speed, therefore, by using the historical running data of the equipment in the healthy state, the baseline curve of the vibration energy of the bearing changing with the rotation speed is fitted by the least square method, and the baseline curve belongs to a one-dimensional quadratic curve without a first-order term, therefore, only the second-order term and the constant term are contained in the fitted baseline curve, the coefficient of the second-order term is taken as the health reference coefficient the constant term is taken as the health reference coefficient , characterizing the proportional coefficient of the vibration energy increasing with the square of the rotation speed, characterizing the inherent noise energy of the system at rest or at very low speed.

[0047] Further, the speed decoupling wear index is calculated by combining the rotation speed signal and the comprehensive impact energy, and the specific calculation formula is as follows:

[0048]

[0049] In the formula: the speed decoupling wear index is represented; the comprehensive impact energy is represented; the rotation speed signal is represented; , the health reference coefficient is represented.

[0050] wherein the numerator term represents the residual extraction, that is, subtracting the theoretically healthy energy at the current rotating speed from the actually measured total energy, so as to peel off the contribution of the speed to the energy, and the residual left is the vibration energy increment additionally generated due to the wear defect; the denominator term represents the scale normalization, which is used to eliminate the influence of the magnitude difference of the reference energy at different rotating speeds: at a high-speed working condition, the reference energy is large, and the same wear increment may account for a very small proportion, and by dividing the residual by the theoretical energy containing the rotating speed term, the residual is standardized; the 1 added in the denominator is to prevent numerical calculation instability when the rotating speed is 0.

[0051] It should be noted that the physical mapping model of the vibration energy changing with the rotating speed is constructed based on the preset healthy reference coefficient, the speed-decoupled wear index removing the influence of the rotating speed is calculated, the index can automatically peel off the energy component naturally increased due to the increase of the rotating speed, thereby completely solving the problems of false positives caused by the energy virtual high due to high-speed operation and weak feature missing caused by low-speed or background noise.

[0052] S4, comparing the speed-decoupled wear index with a preset adaptive alarm threshold, and determining the wear state of the cam follower bearing according to the comparison result.

[0053] It should be noted that after the processing of step S3, the speed-decoupled wear index obtained no longer fluctuates with the rotating speed, but presents a stable curve, which makes it possible to use a single fixed threshold for full-speed domain monitoring, greatly simplifying the control logic and improving the robustness of the system.

[0054] Specifically, the adaptive alarm threshold is set; the speed-decoupled wear index is calculated in real time and compared with the adaptive alarm threshold :

[0055] If , it is determined that the cam follower bearing is in a healthy state; at this time, even if the comprehensive impact energy is large at the high-speed stage, the decoupled index will still tend to 0, thereby avoiding false positives caused by virtual high energy.

[0056] If , it is determined that the cam follower bearing has a wear abnormality; at this time, the system issues an audible and visual alarm, and feeds back the abnormal signal to the PLC controller, and the control equipment performs automatic speed reduction or shutdown operation to prevent mechanical seizure accidents.

[0057] wherein the adaptive alarm threshold The setting follows the closed-loop process of data cleaning, distribution diagnosis and parameter optimization: first, in order to ensure the representativeness and purity of the benchmark data, the historical records of nearly one to three months are extracted, and the non-business noise caused by equipment maintenance, sensor failure or collection error is pre-eliminated, and the speed decoupling wear index of each benchmark data is calculated; then, by drawing the histogram of the speed decoupling wear index of all benchmark data, the distribution form is visually determined: if the histogram presents approximate normal symmetric distribution, the absolute median deviation (MAD) method with stronger robustness is selected to determine the candidate threshold to avoid the pulling of statistical characteristics by accidental outliers, if the histogram presents obvious long tail or skew distribution, the quantile method is directly used to determine the candidate threshold; finally, the obtained candidate threshold is substituted into the benchmark data stream for simulation running, the false alarm rate of the known normal period and the sensitivity of the fault period are counted, according to the false alarm rate and the sensitivity, the coefficients in the absolute median deviation method or the quantile method are dynamically adjusted, so as to adjust the obtained candidate threshold, until the best balance point is found between the false alarm rate and the sensitivity, so as to take the corresponding candidate threshold as the adaptive alarm threshold .

[0058] Exemplarily, Figure 3 The schematic diagram of the prior art for bearing monitoring and early warning through traditional energy monitoring indicators; wherein the curve corresponding to the health state energy represents the energy curve of the bearing in the health state, which is greatly uplifted in the middle high-speed region due to the influence of physical law, and the peak value far exceeds the fixed alarm threshold, therefore, under the prior art, the system will mistakenly determine the bearing failure only because the equipment runs fast, resulting in serious false alarm; the curve corresponding to the wear state energy represents the energy curve of the bearing when there is early wear, although the overall energy is higher than that in the health state, but in the low-speed region at both ends, the energy amplitude is obviously lower than the fixed alarm threshold, therefore, under the prior art, when the equipment is in the low-speed running stage, even if the bearing has been worn, the system cannot detect it, resulting in dangerous false negative.

[0059] Exemplarily, Figure 4The schematic diagram of the bearing monitoring and early warning by the speed decoupling wear index of the application; wherein, for the curve corresponding to the health state index, after the decoupling operation of the application, the interference of the speed fluctuation on the energy is completely eliminated, no matter whether the equipment is in acceleration, deceleration or high speed operation, the health index is always suppressed near the 0 value, and presents a straight line close to the X axis, and is always below the alarm threshold, which proves that the application eliminates the false alarm problem under high speed working condition; for the curve corresponding to the wear state index, after the decoupling operation, the wear characteristics are effectively extracted and normalized, the curve presents a straight suspended line, and the value is stable at about 0.6, it is worth noting that in the low speed area at both ends, the curve is still stably suspended above the green threshold line, and does not fall with the speed, which proves that the application can maintain high sensitivity to wear characteristics in the whole speed domain, and eliminates the missing alarm problem under low speed working condition.

[0060] In summary, by constructing the speed decoupling wear index irrelevant to the speed, the application successfully separates the health and wear states mixed together on the vertical axis, and realizes the precise monitoring effect of the fixed threshold covering the whole working condition.

[0061] The embodiment of the application further discloses a cam follower bearing wear degree monitoring system, comprising a processor and a memory, and the memory stores computer program instructions, when the computer program instructions are executed by the processor, a cam follower bearing wear degree monitoring method according to the application is realized.

[0062] The above system further comprises a communication bus and a communication interface and other components familiar to those skilled in the art, the settings and functions of which are known in the art, and thus will not be repeated here.

Claims

1. A cam follower bearing wear extent monitoring method, characterized by, The method comprises the steps of: synchronously collecting original vibration signals and rotation speed signals of a cam follower bearing, and using the rotation speed signals to perform equal-angle resampling processing on the vibration signals to obtain angular domain vibration signals; using a variational mode decomposition algorithm to decompose the angular domain vibration signals into a plurality of modal components, evaluating impact significance weights of the modal components according to kurtosis values of the modal components and correlation coefficients of the modal components and the original vibration signals, and performing energy-weighted summation of the modal components based on the impact significance weights to obtain comprehensive impact energy; based on a preset health reference coefficient, constructing a physical mapping model for calculating vibration energy using the rotation speed signals; calculating a speed decoupled wear index removing the influence of rotational speed using the integrated impact energy and the vibration energy calculated from the physical mapping model , and , denotes the integrated impact energy, denotes the rotational speed signal, , denotes the health reference coefficient; the health reference coefficient is obtained by: Utilize the historical running data of the equipment in the healthy state, fit the benchmark curve of the bearing vibration energy changing with the speed by the least square method, and the benchmark curve belongs to the one-dimensional quadratic curve not containing the first item, only the quadratic item and the constant item are contained in the benchmark curve obtained by fitting, the coefficient of the quadratic item obtained is taken as the health benchmark coefficient the constant item obtained is taken as the health benchmark coefficient , characterize the proportional coefficient of the vibration energy increasing with the square of the speed, characterize the inherent bottom noise energy of the system at rest or at very low speed; comparing the speed decoupling wear index with a preset adaptive alarm threshold, and determining a wear state of the cam follower bearing according to a comparison result.

2. A cam follower bearing wear extent monitoring method according to claim 1, characterised in that, The calculation formula of the impact significance weight of the modal component is: ； In the formula: Indicates the first Impact significance weights for each modal component; , They represent the first The and the first One modal component; , They represent the first The and the first kurtosis values ​​of each modal component; Represents the original vibration signal; The mode number parameter in the variational mode decomposition algorithm is equal to the number of all modal components obtained by decomposition. ; , Indicates the first The Pearson correlation coefficient between each modal component and the original vibration signal; Indicates the first The Pearson correlation coefficient between each modal component and the original vibration signal; This indicates taking the absolute value.

3. A cam follower bearing wear extent monitoring method according to claim 1, characterized in that, The calculation formula of the comprehensive impact energy is: ； In the formula, denotes the comprehensive impact energy; is a mode number parameter in the variational mode decomposition algorithm, and the number of all mode components obtained by decomposition is equal to ; denotes the impact significance weight of the th mode component; denotes the length of the angular domain vibration signal, and all mode components are equal to ; denotes the th discrete signal point of the th mode component.

4. A cam follower bearing wear extent monitoring method according to claim 1, characterized in that, The original vibration signals are collected by a high-frequency piezoelectric vibration sensor, and the rotation speed signals are synchronously collected through an encoder interface of an industrial field bus or a servo driver.

5. A cam follower bearing wear extent monitoring method according to claim 1, characterized in that, The equal-angle resampling processing of the vibration signals using the rotation speed signals to obtain the angular domain vibration signals comprises the steps of: synchronously collecting vibration time series and corresponding rotation speed time series in a period of time; performing detrending processing on the vibration time series to eliminate direct current components and baseline drift, performing equal-angle resampling on the vibration time series using real-time rotation speed time series, converting time domain non-stationary signals into angular domain stationary signals, and obtaining angular domain vibration signals.

6. A cam follower bearing wear extent monitoring method according to claim 1, characterized in that, The determination of the wear state of the cam follower bearing according to the comparison result comprises the steps of: Real-time calculation of speed decoupled wear index and compared to adaptive alarm thresholds : if the speed decoupling wear index is less than or equal to the adaptive alarm threshold, determining that the cam follower bearing is in a healthy state; if the speed decoupling wear index is greater than the adaptive alarm threshold, determining that the cam follower bearing has wear abnormality and triggering an alarm signal.

7. A cam follower bearing wear extent monitoring method according to claim 1, characterized in that, The setting method of the adaptive alarm threshold comprises the steps of: extracting historical records in the past one to three months and eliminating non-business noise points, and calculating speed decoupling wear indexes of each reference data obtained; drawing a histogram of the speed decoupling wear indexes of all reference data to determine the distribution form:

8. A cam follower bearing wear extent monitoring system, characterized by, if the histogram presents an approximately normal symmetric distribution, using an absolute median difference method to determine a candidate threshold, and if the histogram presents a significant long tail or skew distribution, using a quantile method to determine a candidate threshold; substituting the obtained candidate threshold into the reference data stream for simulation running, counting a false alarm rate in a known normal period and a sensitivity in a fault period, dynamically adjusting coefficients in the absolute median difference method or the quantile method according to the false alarm rate and the sensitivity, thereby adjusting the obtained candidate threshold, and finding a best balance point between the false alarm rate and the sensitivity, and taking the corresponding candidate threshold as the adaptive alarm threshold. The method comprises: a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, a cam follower bearing wear degree monitoring method according to any one of claims 1-7 is realized.

Images