CN-121977840-A - Method and device for identifying early abnormal state of thrust bearing

CN121977840ACN 121977840 ACN121977840 ACN 121977840ACN-121977840-A

Abstract

The disclosure belongs to the technical field of nuclear power, and particularly relates to a method and a device for identifying early abnormal states of a thrust bearing. By means of the normal state storage characteristics and the strategy of the abnormal state storage original data, on the premise that key information is not lost, storage pressure caused by massive acoustic emission data is greatly relieved, and system deployment and operation and maintenance costs are reduced. The method constructs a complete closed loop of initial verification, reference establishment, real-time monitoring, intelligent diagnosis, data retention. Through comprehensively analyzing four key characteristics of high-frequency, medium-frequency, low-frequency energy and effective values of background noise, the fault types can be primarily distinguished, such as poor lubrication and overload, and a more accurate basis is provided for maintenance decision.

Inventors

Hou Xiuqun
MIAO BIQI
ZHANG ZHAOGUANG
TANG MINGJUN
HU DI
ZHANG YIZHOU
JIANG QINGLEI

Assignees

中核武汉核电运行技术股份有限公司

Dates

Publication Date: 20260505
Application Date: 20251125

Claims (10)

1. A method for identifying an early abnormal state of a thrust bearing, the method comprising: Step 1, acquiring acoustic emission signals of a newly installed thrust bearing, and comprehensively judging whether the bearing is in a normal state by combining the temperature of lubricating oil, the acoustic signals and historical normal data; Step 2, acquiring acoustic emission signals in a normal state, splitting the acquired acoustic emission signals into M data fragments, and determining characteristic parameters of each fragment, wherein the characteristic parameters comprise high-frequency band energy, intermediate-frequency band energy, low-frequency band energy and background noise effective values; Step 3, respectively determining the mean value X and the standard deviation Y of four characteristic parameters of the M data fragments, and setting the normal threshold value of each characteristic parameter by X+/-N multiplied by Y based on a Gaussian distribution principle; step4, determining characteristic parameters of the currently acquired acoustic emission signals, and comparing the characteristic parameters of the currently acquired acoustic emission signals with normal thresholds of the characteristic parameters; And 5, judging that the thrust bearing is poor in lubrication when the low-frequency energy exceeds a threshold value or the low-frequency energy and the high-frequency energy exceed the threshold value simultaneously, and judging that the thrust bearing is overloaded when the effective value of the noise exceeds the threshold value.
2. The method according to claim 1, wherein in the step 2, the number M of the data segments is greater than 30, and the length of a single data segment is not less than 0.1 seconds.
3. The method according to claim 1, wherein in step 3, the coefficient N is adjustable according to the on-site monitoring requirements.
4. The method of claim 1, further comprising the step of data storage: Under the normal running state, a section of acoustic emission original data is stored at a first frequency, and the rest time only stores the characteristic parameters; And under an abnormal operation state, storing acoustic emission original data fragments at a second frequency higher than the first frequency on the basis of storing the characteristic parameters, wherein the abnormal operation state is triggered by mutation of the characteristic parameters.
5. The method according to claim 1, wherein the background noise effective value is obtained by: sliding the acoustic emission signal sequence with the overlapping rate epsilon by adopting a sliding window with the length of N; calculating the effective value of the signal in each sliding window to form an effective value sequence; dividing the effective value sequence from the minimum value to the maximum value into M equal parts and drawing a distribution histogram; And taking an abscissa value corresponding to the first highest point in the distribution histogram as the background noise effective value.
6. The method of claim 5, wherein the length N of the sliding window is determined by: Initializing N to be the minimum value of the length L of the impact signal and the length S of the 0.25 times of the background noise signal; sliding a window in a signal section only containing the background noise, and calculating the variance of the obtained background noise characteristic value sequence; If the variance exceeds the set value omega, increasing the window length N, and repeating the steps until the variance is smaller than or equal to omega.
7. The method of claim 5, wherein epsilon is adjusted based on the smoothness of the plotted distribution histogram and the computational resource profile by increasing epsilon if the smoothness of the histogram is less than a predetermined threshold, by decreasing epsilon if the smoothness is greater than a predetermined threshold, or by tightening the computational resource.
8. An early abnormal state recognition device for a thrust bearing, the device comprising: The initial state verification module is used for collecting acoustic emission signals of the newly installed thrust bearing and comprehensively judging whether the bearing is in a normal state or not by combining the temperature of lubricating oil, the acoustic signals and historical normal data; the health reference construction module is used for collecting acoustic emission signals in a normal state, splitting the collected acoustic emission signals into M data fragments, and determining characteristic parameters of each fragment, wherein the characteristic parameters comprise high-frequency band energy, middle-frequency band energy, low-frequency band energy and background noise effective values; The threshold value determining module is used for respectively determining the mean value X and the standard deviation Y of four characteristic parameters of the M data fragments, and setting the normal threshold value of each characteristic parameter by X+/-N multiplied by Y based on the Gaussian distribution principle; the real-time monitoring analysis module is used for determining the characteristic parameters of the currently acquired acoustic emission signals and comparing the characteristic parameters of the currently acquired acoustic emission signals with the normal threshold values of the characteristic parameters; the abnormal state diagnosis module is used for judging that the thrust bearing is poor in lubrication when the low-frequency energy exceeds the threshold value or the low-frequency energy and the high-frequency energy exceed the threshold value simultaneously, and judging that the thrust bearing is overloaded when the background noise effective value exceeds the threshold value.
9. An early abnormal state recognition device for a thrust bearing, the device comprising: A processor; A memory for storing processor-executable instructions; Wherein the processor is configured to perform the method of any one of claims 1 to 7.
10. A non-transitory computer readable storage medium having stored thereon computer program instructions, which when executed by a processor, implement the method of any of claims 1 to 7.

Description

Method and device for identifying early abnormal state of thrust bearing Technical Field The invention belongs to the field of intelligent monitoring and diagnosis of rotary machinery, and particularly relates to a method and a device for identifying early abnormal states of a thrust bearing. Background The thrust bearing is widely applied to key equipment such as steam turbines, pumps and the like, the conventional temperature monitoring method for the thrust bearing cannot timely identify early abnormal states such as bearing load change, poor lubrication and the like, so that the problem of burning of the bearing occurs under extreme working conditions, and in view of the situation, the early abnormal states of the bearing are required to be effectively identified, and related measures are taken to avoid the damage of the thrust bearing. Disclosure of Invention In order to overcome the problems in the related art, the method and the device for identifying the early abnormal state of the thrust bearing are provided. According to an aspect of the embodiments of the present disclosure, there is provided a method for identifying an early abnormal state of a thrust bearing, the method including: Step 1, acquiring acoustic emission signals of a newly installed thrust bearing, and comprehensively judging whether the bearing is in a normal state by combining the temperature of lubricating oil, the acoustic signals and historical normal data; Step 2, acquiring acoustic emission signals in a normal state, splitting the acquired acoustic emission signals into M data fragments, and determining characteristic parameters of each fragment, wherein the characteristic parameters comprise high-frequency band energy, intermediate-frequency band energy, low-frequency band energy and background noise effective values; Step 3, respectively determining the mean value X and the standard deviation Y of four characteristic parameters of the M data fragments, and setting the normal threshold value of each characteristic parameter by X+/-N multiplied by Y based on a Gaussian distribution principle; step4, determining characteristic parameters of the currently acquired acoustic emission signals, and comparing the characteristic parameters of the currently acquired acoustic emission signals with normal thresholds of the characteristic parameters; And 5, judging that the thrust bearing is poor in lubrication when the low-frequency energy exceeds a threshold value or the low-frequency energy and the high-frequency energy exceed the threshold value simultaneously, and judging that the thrust bearing is overloaded when the effective value of the noise exceeds the threshold value. In a possible implementation manner, in the step 2, the number M of the data segments is greater than 30, and the length of the single data segment is not less than 0.1 seconds. In a possible implementation manner, in the step 3, the coefficient N can be adjusted according to the on-site monitoring requirement. In one possible implementation, the method further includes a data storage step: Under the normal running state, a section of acoustic emission original data is stored at a first frequency, and the rest time only stores the characteristic parameters; And under an abnormal operation state, storing acoustic emission original data fragments at a second frequency higher than the first frequency on the basis of storing the characteristic parameters, wherein the abnormal operation state is triggered by mutation of the characteristic parameters. In one possible implementation, the background noise effective value is obtained by: sliding the acoustic emission signal sequence with the overlapping rate epsilon by adopting a sliding window with the length of N; calculating the effective value of the signal in each sliding window to form an effective value sequence; dividing the effective value sequence from the minimum value to the maximum value into M equal parts and drawing a distribution histogram; And taking an abscissa value corresponding to the first highest point in the distribution histogram as the background noise effective value. In one possible implementation, the length N of the sliding window is determined by: Initializing N to be the minimum value of the length L of the impact signal and the length S of the 0.25 times of the background noise signal; sliding a window in a signal section only containing the background noise, and calculating the variance of the obtained background noise characteristic value sequence; If the variance exceeds the set value omega, increasing the window length N, and repeating the steps until the variance is smaller than or equal to omega. In one possible implementation, epsilon is adjusted according to the smoothness of the plotted distribution histogram and the computational resource situation, epsilon is increased if the smoothness of the histogram is less than a preset threshold, epsilon is decreased if the smoothness is