Turbine shaft seal friction fault feature detection method and system based on vibration signals
By installing vibration sensors and key phase sensors on the turbine, the vibration signals at high and low speeds are processed, impact characteristics are extracted and singular energy calculations are strengthened, and the problem of difficulty in detecting shaft seal friction failures during high-speed operation of the turbine is solved, achieving accurate detection of early faults and reducing safety risks.
Patent Information
- Application Number
- CN202411816572.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-11
- Publication Date
- 2025-05-23
AI Technical Summary
The prior art is difficult to effectively detect shaft seal friction failures when the turbine is running at high speed, resulting in failure to detect early failures in time, which in turn causes middle and late failures and safety risks.
By installing a vibration sensor and a key phase sensor, the vibration signal of the turbine at high and low speeds is tested, and the signal processing is performed using the key phase pulse as a reference, the smoothness interference of the rotating shaft surface is removed, the impact characteristics in the signal are extracted, and fault characteristic detection is strengthened through the singular energy calculation method.
Accurate detection of early friction faults is achieved, the accuracy of diagnosis is improved, the friction faults are avoided to develop in the middle and late stages, and the safety risks are reduced.
Smart Images
Figure CN120028050A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to a method and a system for detecting characteristics of a steam turbine shaft seal friction fault, and in particular to a method and a system for detecting characteristics of a steam turbine shaft seal friction fault based on a vibration signal. Background Art
[0002] Large rotating machinery such as steam turbines are composed of rotating parts and stationary parts, with a gap designed between the two. In order to improve economy, it is necessary to reduce the leakage of working media such as steam in the steam turbine to the low-pressure area, so the dynamic and static gaps are mostly designed to be relatively small. Affected by factors such as unit installation deviation, operating parameter deviation and large vibration, friction is prone to occur between the dynamic and static parts. Friction can cause shaft seal wear, cause unit vibration fluctuations and divergence, and further cause large shaft bending accidents, which have a great impact on the safe and economical operation of the unit.
[0003] It is of great significance to study the fault characteristic detection method in the early stage of friction fault, so as to discover the friction fault as early as possible and avoid the development of friction fault to the middle and late stages.
[0004] At present, the friction monitoring methods mainly include: monitoring methods based on vibration characteristics and monitoring methods based on acoustic emission characteristics.
[0005] When the turbine speed is relatively low (for example, below 500r / min), the steam flow noise is small, and the friction noise can be detected in this way. Therefore, the turbine operation procedures are generally designed with a low-speed warm-up condition, and the friction noise needs to be monitored at this speed. When the speed increases, the steam flow noise in the cylinder is large, and the acoustic emission signal generated by friction is basically completely covered by the steam flow noise. The acoustic emission diagnostic technology is basically ineffective at high speed and rated speed of the turbine. Current research mainly focuses on how to eliminate the influence of steam flow noise.
[0006] Friction will generate impact force on the shaft. Under the action of this force, impact, distortion and other phenomena will appear in the vibration signal of the shaft. In order to reduce the risk of shaft seal friction and reduce its harm, the tooth tips of turbine shaft seal teeth are generally processed to be sharper. The friction force is small. Characteristics such as impact caused by friction are masked by the original vibration signal. Friction fault diagnosis is mostly based on the fluctuation and rise of the rotation frequency component in the vibration signal. However, at this time, the friction fault has entered the middle and late stages, and the risk of serious accidents such as large shaft bending is relatively high. It is necessary to study the friction fault feature enhancement technology to improve the accuracy of early friction fault diagnosis. Summary of the invention
[0007] Purpose of the invention: The purpose of the present invention is to provide a method and system for detecting the characteristics of turbine shaft seal friction faults based on vibration signals, thereby improving the accuracy of early friction fault diagnosis by enhancing the impact characteristics in the vibration signals caused by friction faults.
[0008] Technical solution: The present invention comprises the following steps:
[0009] Sensor layout: Install vibration sensors and key phase sensors. The vibration sensors are arranged at the friction monitoring part of the steam turbine. Key grooves are set on the surface of the rotating shaft.
[0010] Test the vibration signal and key phase signal continuously output by the sensor in several rotation cycles: Test the vibration signal of the steam turbine at high speed and low speed respectively, and take the value at equal angle intervals based on the key phase pulse, subtract the high and low speed equal interval sampling signals, and remove the interference of the surface finish of the shaft;
[0011] Taking the key phase pulse signal as a reference, the vibration waveform Δx(i), i=1,2,...,N is split into M segments, each segment contains K pulses, the vibration waveforms in the M segments are added and averaged to obtain the vibration signal averaged between the K key phase pulses;
[0012] Define the calculation method of the singular energy of the signal: the singular energy of a smooth harmonic signal is 0, and an impact signal will produce a larger onset volume at the singular point.
[0013] When the key slot passes through the key phase sensor, the key phase sensor generates a pulse output, and the pulse period is completely equal to the rotation period.
[0014] The test of the vibration signal and key phase signal continuously output by the sensor during several rotation cycles specifically includes:
[0015] Record the sensor output signal and key phase signal at low speed, and record several cycles continuously, which are recorded as x 0 (t),y 0 (t);
[0016] Taking the key phase signal as the reference, the vibration signal between two key phase pulses is divided into a number of N segments with equal intervals, and the vibration value of each point is obtained, which is recorded as: x 0 (i), i = 1, ..., N, which is used as the axis surface finish error benchmark;
[0017] When rotating at high speed, the vibration signal is tested continuously in several rotation cycles according to the above method, recorded as x 1 (t), taking the key phase signal as the reference, the vibration signal between two key phase pulses is divided into several equally spaced segments, and the vibration value of each point is obtained, which is recorded as: x 1 (i), i=1,2,...,N;
[0018] Solve to get the difference Δx(i) of the signal waveform at high speed and low speed, i=1,2,...,N:
[0019] Δx(i)=x 1 (i)-x0 (i).
[0020] The average vibration signal between the K key phase pulses is:
[0021] The method for calculating the singular energy of the defined signal is specifically as follows:
[0022] Introduce the signal singular energy evaluation operator: In the formula, M[·] represents the mean value of the variable;
[0023] For smooth signals E(t) = 0, indicating that there are no singular points in the waveform within the entire signal analysis interval;
[0024] For discrete signals, the above formula can be expressed as:
[0025] E n =(x n 2 -x n-1 x n+1 )-(M[x n 2 -x n-1 x n+1 ]) 2 .
[0026] The singular energy distribution of each point on the signal waveform is calculated through three adjacent points on the signal waveform.
[0027] The singular point on the signal waveform is a fault feature.
[0028] The vibration sensor and the key phase sensor are both facing the surface of the rotating shaft.
[0029] The vibration sensor and the key phase sensor are both connected to the dynamic signal analyzer.
[0030] A steam turbine shaft seal friction fault feature detection system based on vibration signals, comprising:
[0031] Sensor layout module: install vibration sensor and key phase sensor; open key slot on the surface of rotating shaft;
[0032] The vibration signal and key phase signal test module of the sensor continuously outputs: the vibration signal of the steam turbine is tested at high speed and low speed respectively, and the key phase pulse is used as the reference to take values at equal angle intervals, and the high and low speed equal interval sampling signals are subtracted to remove the interference of the surface finish of the shaft;
[0033] Averaged vibration signal acquisition module: Based on the key phase pulse signal, the vibration waveform is divided into M segments, each segment contains K pulses, the vibration waveforms in the M segments are added and averaged to obtain the averaged vibration signal between the K key phase pulses;
[0034] Signal singular energy evaluation module: introduces the signal singular energy evaluation operator to calculate the singular energy distribution of each point on the signal waveform.
[0035] Beneficial effects: The present invention realizes the monitoring and diagnosis of early friction faults by strengthening the impact characteristics in the vibration signal caused by friction faults; by conducting research on fault characteristic detection methods in the early stage of friction faults, friction faults can be discovered as early as possible to avoid the development of friction faults to the middle and late stages, which is of great significance. BRIEF DESCRIPTION OF THE DRAWINGS
[0036] Figure 1 It is a schematic diagram of the sensor layout of the present invention;
[0037] Figure 2 for Figure 1 Key phase signal diagram in ;
[0038] Figure 3 It is the vibration signal output by the sensor within one rotation cycle at low speed;
[0039] Figure 4 It is the key phase signal output by the sensor within one rotation cycle at low speed;
[0040] Figure 5 It is the vibration signal continuously output by the sensor within several rotation cycles;
[0041] Figure 6 It is the key phase signal continuously output by the sensor within several rotation cycles;
[0042] Figure 7 To compensate for the vibration signal after the surface finish error of the rotating shaft;
[0043] Figure 8 To compensate the key phase signal after the surface finish error of the rotating shaft;
[0044] Fig. 9 The vibration signal and the key phase signal are divided into several points collected at equal intervals based on the key phase signal;
[0045] Fig.10 is the vibration signal before and after averaging;
[0046] Fig.11 It is the distribution diagram of singular points on the signal waveform. DETAILED DESCRIPTION
[0047] The present invention will be further described below in conjunction with the accompanying drawings.
[0048] Example 1
[0049] The method for detecting the characteristics of the turbine shaft seal friction fault based on the vibration signal of this embodiment comprises the following steps:
[0050] S1. Sensor layout: Install vibration sensor 1 at the friction monitoring part near the turbine shaft seal, and the vibration sensor 1 faces the surface of the rotating shaft; install key phase sensor 3, and the key phase sensor 3 faces the surface of the rotating shaft. Both vibration sensor 1 and key phase sensor 3 are connected to the dynamic signal analyzer, and the vibration displacement signal and key phase signal during the rotation of the rotating shaft are tested by the dynamic signal analyzer; a key groove 2 is opened on the surface of the rotating shaft, such as Figure 1 shown.
[0051] When the machine is working, the shaft rotates. Figure 2 As shown, when the key slot 2 passes through the key phase sensor 3, the key phase sensor 3 will generate a pulse output, and the pulse period is completely equal to the rotation period.
[0052] S2. Test the vibration signal and key phase signal continuously output by the sensor in several rotation cycles: According to the characteristics of turbine shaft seal friction failure, test the vibration signal of the turbine at high speed and low speed respectively, and take the value at equal angle intervals based on the key phase pulse. The signal obtained by subtracting the high and low speed equal interval sampling signals can remove the interference of the surface finish of the rotating shaft. Specifically include:
[0053] S21, record the sensor output signal and key phase signal at low speed, and record several cycles continuously, which are recorded as x 0 (t),y 0 (t);
[0054] S22, taking the key phase signal as a reference, divide the vibration signal between two key phase pulses into a number of N segments with equal intervals, and obtain the vibration value of each point, which is recorded as: x 0 (i), i = 1, ..., N, which is used as the axis surface finish error benchmark;
[0055] S23, when rotating at high speed, continuously test the vibration signal within several rotation cycles according to the above method, recorded as x 1 (t), taking the key phase signal as the reference, the vibration signal between two key phase pulses is divided into several equally spaced segments, and the vibration value of each point is obtained, which is recorded as: x 1 (i), i=1,2,...,N;
[0056] S3. Solve to obtain the difference Δx(i) of the signal waveform at high speed and low speed, i=1,2,...,N:
[0057] Δx(i)=x1 (i)-x 0 (i)
[0058] The signal waveform difference Δx(i) obtained after this processing eliminates the influence of the surface finish of the shaft on the vibration and can more realistically reflect the vibration condition of the equipment.
[0059] S4. Based on the key phase pulse signal, the vibration waveform Δx(i), i=1,2,...,N is divided into M segments, each of which contains K pulses. The vibration waveforms in the M segments are added and averaged to obtain the average vibration signal between the K key phase pulses:
[0060]
[0061] When there is dynamic and static friction between the shaft and the shaft seal, the shaft will be impacted by force, and impact features will appear in the vibration signal. However, because the friction force is weak, the impact features in the vibration signal are mostly masked by normal vibration signals and noise signals. In the early friction stage, the vibration impact features caused by friction are even weaker, and it is difficult to use the impact features to diagnose friction faults.
[0062] The gap between the moving and static parts near the high point of vibration on the rotating shaft is the smallest, and the arc segment near the high point is most likely to rub against the shaft seal. Therefore, the impact signal caused by friction has a relatively strong periodicity, and its period is the same as the turbine rotation period. Compared with the impact signal, the noise signal does not have periodicity. Using this periodic feature, the noise signal can be removed.
[0063] S5. Define the calculation method of the singular energy of the signal: The singular energy of the smooth harmonic signal is 0, and the impact signal will produce a large volume at the singular point. According to the size of the singular energy and the periodic characteristics of the distribution of the singular points, the enhanced detection of the impact characteristics of the early friction fault can be achieved, and the accuracy of the early friction fault diagnosis can be improved.
[0064] Introduce the signal singular energy evaluation operator: In the formula, M[·] represents the mean value of the variable.
[0065] For smooth signals E(t) = 0, indicating that there are no singular points in the waveform within the entire signal analysis interval;
[0066] For discrete signals, the above formula can be expressed as:
[0067]
[0068] The singular energy distribution of each point on the signal waveform can be calculated through three adjacent points on the signal waveform. Most of the singular points on the waveform are related to faults and are fault characteristics.
[0069] Through the above signal processing method, it is possible to achieve enhanced detection of periodic impact features hidden in the signal waveform, and diagnose friction faults based on the distribution characteristics of periodic singular points appearing on the waveform.
[0070] The present embodiment is described in detail below with reference to the accompanying drawings:
[0071] Step 1: If Figure 1 As shown, a vibration sensor 1 is installed at the friction monitoring part near the turbine shaft seal, and the sensor faces the shaft surface. A key slot 2 is opened on the shaft surface; a set of key phase sensors 3 are installed, and the key phase sensors 3 face the shaft surface. The vibration displacement signal and key phase signal during the shaft rotation process are tested by a dynamic signal analyzer.
[0072] When the turbine rotates, the key phase sensor 3 outputs a series of equally spaced pulse signals, and the pulse period is the same as the rotation period, such as Figure 2 As shown, this is used as the basis for full-cycle sampling of the signal and error elimination.
[0073] Step 2: Test the vibration signal and key phase signal output by the sensor within one rotation cycle at low speed, such as Figure 3 and Figure 4 As shown, this is used as the shaft surface finish interference signal, denoted as x 0 (t),y 0 (t). In the figure, the number of cycles is 2, the rotation frequency is f = 1 Hz, and the number of sampling points per cycle is 2048. The discrete sampling points of the vibration signal between the two key phase pulse signals are denoted as x 0 (i).
[0074] Step 3: Test the vibration signal and key phase signal continuously output by the sensor in several rotation cycles at high speed, such as Figure 5 and Figure 6 As shown in the figure, the rotation frequency f = 50Hz, the total number of sampling cycles is 128, and the number of sampling points per cycle is 2048. Taking the key phase signal as the reference, the vibration signal between two key phase pulses is divided into several segments with equal intervals, and the vibration value of each point is obtained, which is recorded as: x 1 (i).
[0075] Step 4: Remove the interference of the shaft surface finish from the original vibration signal. Affected by the machining accuracy, there are burrs, ovality, pits and other phenomena on the shaft surface, which will interfere with the output signal and affect the accuracy and reliability of the feature extraction algorithm. The shaft vibration is very small at low speeds and can be approximately regarded as no vibration. Therefore, the output signal of the vibration sensor at low speed can be regarded as the interference signal of the shaft surface finish, which needs to be eliminated from the measured vibration signal at high speed rotation. Calculate the difference Δx(i) of the signal waveform at high and low speeds:
[0076] Δx(i)=x 1 (i)-x 0 (i)
[0077] The signal Δx(i) obtained after this processing is as follows: Figure 7 and Figure 8 As shown, the influence of the shaft surface finish on the vibration is eliminated, and the vibration condition of the equipment can be reflected more realistically.
[0078] Step 5: If Fig. 9 As shown, based on the key phase pulse signal, the vibration waveform Δx(i), i=1,2,...,N is split into M segments, each of which contains K pulses.
[0079] The vibration waveforms in M segments are added and averaged to obtain the average vibration signal between K key phase pulses, which eliminates the influence of noise.
[0080]
[0081] Fig.10 The vibration signals before and after averaging are given. In the figure, the total sampling period is 64, the period in each segment is 4, and the total number of segments is 16.
[0082] Step 6: Most of the singular points on the waveform are related to faults and are fault characteristics. Introduce the signal singular energy evaluation operator:
[0083]
[0084] In the formula, M[·] represents the mean value of the variable.
[0085] For smooth signals For example, E(t) = 0, which means that there are no singular points in the waveform within the entire signal analysis interval.
[0086] For discrete signals, the above formula can be expressed as:
[0087] E n =(x n 2 -x n-1 x n+1 )-(M[x n 2 -x n-1 x n+1 ]) 2
[0088] Through the three adjacent points on the signal waveform, the singular energy distribution of each point on the signal waveform can be calculated, such as Fig.11 shown.
[0089] Example 2
[0090] The steam turbine shaft seal friction fault feature detection system based on vibration signals of this embodiment includes:
[0091] Sensor layout module: install a vibration sensor at the friction monitoring position near the turbine shaft seal, with the vibration sensor facing the surface of the rotating shaft; install a key phase sensor, with the key phase sensor facing the surface of the rotating shaft. Both the vibration sensor and the key phase sensor are connected to a dynamic signal analyzer, and the vibration displacement signal and key phase signal during the rotation of the rotating shaft are tested by the dynamic signal analyzer; a key slot is opened on the surface of the rotating shaft.
[0092] The vibration signal and key phase signal test module of the sensor continuously outputs: According to the characteristics of turbine shaft seal friction failure, the vibration signal of the turbine is tested at high and low speeds respectively, and the key phase pulse is used as the reference to take values at equal angle intervals. The signal obtained by subtracting the high and low speed equally spaced sampling signals can remove the interference of the surface finish of the shaft.
[0093] Signal waveform difference solving module: The signal waveform difference at high speed and low speed is solved. The signal waveform difference obtained after this processing eliminates the influence of the surface finish of the shaft on the vibration, and can more realistically reflect the vibration of the equipment.
[0094] Averaged vibration signal acquisition module: Based on the key phase pulse signal, the vibration waveform is divided into M segments, each segment contains K pulses, the vibration waveforms in the M segments are added and averaged to obtain the averaged vibration signal between the K key phase pulses, which eliminates the influence of noise.
[0095] Signal singular energy evaluation module: The signal singular energy evaluation operator is introduced. Through three adjacent points on the signal waveform, the singular energy distribution of each point on the signal waveform can be calculated.
Claims
1. A method for detecting characteristics of turbine shaft seal friction fault based on vibration signals, characterized in that: The following steps are involved: Sensor layout: Install vibration sensors and key phase sensors. The vibration sensors are arranged at the friction monitoring part of the steam turbine. Key grooves are set on the surface of the rotating shaft. Test the vibration signal and key phase signal continuously output by the sensor in several rotation cycles: Test the vibration signal of the steam turbine at high speed and low speed respectively, and take the value at equal angle intervals based on the key phase pulse, subtract the high and low speed equal interval sampling signals, and remove the interference of the surface finish of the shaft; Taking the key phase pulse signal as a reference, the vibration waveform Δx(i), i=1,2,...,N is split into M segments, each segment contains K pulses, the vibration waveforms in the M segments are added and averaged to obtain the vibration signal averaged between the K key phase pulses; Define the calculation method of the singular energy of the signal: the singular energy of a smooth harmonic signal is 0, and an impact signal will produce a larger onset volume at the singular point.
2. The method for detecting characteristics of turbine shaft seal friction faults based on vibration signals according to claim 1, characterized in that: When the key slot passes through the key phase sensor, the key phase sensor generates a pulse output, and the pulse period is completely equal to the rotation period.
3. The method for detecting characteristics of turbine shaft seal friction faults based on vibration signals according to claim 1, characterized in that: The test of the vibration signal and key phase signal continuously output by the sensor during several rotation cycles specifically includes: Record the sensor output signal and key phase signal at low speed, and record several cycles continuously, which are recorded as x0(t) and y0(t) respectively; Taking the key phase signal as the reference, the vibration signal between two key phase pulses is divided into a number of N segments with equal intervals, and the vibration value of each point is obtained, which is recorded as: x0(i), i=1,...,N, and used as the reference of the surface finish error of the shaft; When rotating at high speed, the vibration signal within several rotation cycles is continuously tested according to the above method, which is recorded as x1(t). The vibration signal between two key phase pulses is divided into several segments with equal intervals based on the key phase signal, and the vibration value of each point is obtained, which is recorded as: x1(i), i=1,2,...,N; Solve to get the difference Δx(i) of the signal waveform at high speed and low speed, i=1,2,...,N: Δx(i)=x1(i)-x0(i).
4. The method for detecting characteristics of turbine shaft seal friction faults based on vibration signals according to claim 3 is characterized in that: The average vibration signal between the K key phase pulses is:
5. The method for detecting characteristics of turbine shaft seal friction faults based on vibration signals according to claim 4, characterized in that: The method for calculating the singular energy of the defined signal is specifically as follows: Introduce the signal singular energy evaluation operator: In the formula, M[·] represents the mean value of the variable; For smooth signals E(t) = 0, indicating that there are no singular points in the waveform within the entire signal analysis interval; For discrete signals, the above formula can be expressed as: E n =(x n 2 -x n-1 x n+1 )-(M[x n 2 -x n-1 x n+1 ]) 2 。 6. The method for detecting characteristics of turbine shaft seal friction faults based on vibration signals according to claim 5, characterized in that: The singular energy distribution of each point on the signal waveform is calculated through three adjacent points on the signal waveform.
7. The method for detecting characteristics of turbine shaft seal friction faults based on vibration signals according to claim 6, characterized in that: The singular point on the signal waveform is a fault feature.
8. The method for detecting characteristics of turbine shaft seal friction faults based on vibration signals according to claim 1, characterized in that: The vibration sensor and the key phase sensor are both facing the surface of the rotating shaft.
9. The method for detecting characteristics of turbine shaft seal friction faults based on vibration signals according to claim 8, characterized in that: The vibration sensor and the key phase sensor are both connected to the dynamic signal analyzer.
10. A steam turbine shaft seal friction fault feature detection system based on vibration signals, comprising: Sensor layout module: install vibration sensor and key phase sensor; open key slot on the surface of rotating shaft; The vibration signal and key phase signal test module of the sensor continuously outputs: the vibration signal of the steam turbine is tested at high speed and low speed respectively, and the key phase pulse is used as the reference to take values at equal angle intervals, and the high and low speed equal interval sampling signals are subtracted to remove the interference of the surface finish of the shaft; Averaged vibration signal acquisition module: Based on the key phase pulse signal, the vibration waveform is divided into M segments, each segment contains K pulses, the vibration waveforms in the M segments are added and averaged to obtain the averaged vibration signal between the K key phase pulses; Signal singular energy evaluation module: introduces the signal singular energy evaluation operator to calculate the singular energy distribution of each point on the signal waveform.