CN-122020035-A - Multi-mode triggered turbine vibration data completeness storage method and related device

CN122020035ACN 122020035 ACN122020035 ACN 122020035ACN-122020035-A

Abstract

The invention discloses a multimode triggered steam turbine vibration data completeness storage method and a related device, belonging to the technical field of steam turbine fault monitoring; the method establishes a real-time monitoring loop for capturing sudden anomalies by comparing the form similarity of the current vibration spectrum form with the form similarity of the dynamic reference, and establishes a reference calibration loop for correcting chronic degradation drift by comparing the form similarity of the dynamic reference with the form similarity of the original health state of the generated spectrum.

Inventors

LI JINKU
Han Baogeng
ZHAO JIXING
LI CHONGCHENG
FANG WEIWEI
LU DAYONG
WANG BO
CAO LI
LIU FENGXIANG
XIAO LISHENG
LI WEI
ZHANG CHAO

Assignees

华能巢湖发电有限责任公司
西安热工研究院有限公司

Dates

Publication Date: 20260512
Application Date: 20260121

Claims (10)

1. A multimode triggered turbine vibration data completeness storage method is characterized by comprising the following steps: Establishing a backtracking buffer area for storing recent historical vibration data; processing vibration signals acquired in real time into a current spectrum morphology vector, and calculating a dynamic reference spectrum morphology vector representing a recent steady state based on recent historical vibration data in the backtracking buffer area; calculating a first morphological similarity between the current spectrum morphological vector and the dynamic reference spectrum morphological vector in real time, and triggering a first storage instruction to permanently store data in the backtracking buffer area when the first morphological similarity is lower than a first preset threshold value; The method comprises the steps of storing an generated spectrum morphology vector representing a reference health state of a steam turbine in advance, and calculating a second morphology similarity between the dynamic reference spectrum morphology vector and the generated spectrum morphology vector; And when the second form similarity is lower than a second preset threshold value, detecting drift caused by chronic degradation, triggering a second storage instruction, and permanently storing the data in the backtracking buffer.
2. The multimode triggered steam turbine vibration data completeness storage method of claim 1, wherein the calculation method of the dynamic reference spectrum morphology vector is as follows: continuously performing short-time Fourier transform on vibration signals acquired in real time to obtain a series of continuous frequency spectrum frames; averaging a plurality of spectrum frames in a preset time period to generate a dynamic reference spectrum morphology vector; The first morphological similarity is determined by calculating the cosine similarity of the current spectral morphology vector and the dynamic reference spectral morphology vector, and the second morphological similarity is determined by calculating the cosine similarity of the dynamic reference spectral morphology vector and the created spectral morphology vector.
3. The method for storing the completeness of vibration data of the steam turbine triggered by multiple modes according to claim 1, wherein when the similarity of the first form is lower than a first preset threshold value, triggering a first storage instruction, and permanently storing the data in the backtracking buffer zone, specifically comprising the following steps: When the first form similarity is lower than the first preset threshold value for the first time, entering a candidate triggering state and starting a confirmation window period; Continuously checking whether the subsequently calculated first form similarity is continuously lower than the first preset threshold value in the confirmation window period; and finally confirming triggering the permanent storage only when the verification is passed, and tracing the triggering time to the time when the first form similarity is lower than a first preset threshold value for the first time.
4. The multi-mode triggered turbine vibration data integrity storage method of claim 1, wherein when said first morphology similarity is below a first preset threshold, triggering a first storage instruction, and permanently storing data in said backtracking buffer, further comprising: calculating a spectrum difference vector between the current spectrum morphology vector causing triggering and the dynamic reference spectrum morphology vector, wherein the calculation formula of the spectrum difference vector is as follows: In the formula, For the current spectral morphology vector, Is a dynamic reference spectrum morphology vector; Is a spectrum difference vector; identifying key frequency features causing the first morphological similarity to be lower than a first preset threshold value by carrying out peak search on the spectrum difference vector; and taking the key frequency characteristic as diagnosis metadata, and permanently storing the key frequency characteristic and the data in the backtracking buffer area.
5. The multi-mode triggered steam turbine vibration data integrity storage method of claim 1, wherein said first predetermined threshold determination method comprises: a health state model is built in advance, the health state model is related to statistical distribution characteristics of first form similarity corresponding to different operation conditions and health states of the steam turbine under respective conditions, and the statistical distribution characteristics comprise mean value and standard deviation of the first form similarity; monitoring the operation condition parameters of the steam turbine in real time; Inquiring and dynamically determining a first preset threshold value in the health state model according to the operation condition parameters monitored in real time.
6. The multi-mode triggered turbine vibration data integrity storage method of claim 1, further comprising: when an operation event that a turbine unit is started and stopped or the load is greatly and rapidly changed is monitored, triggering a third storage instruction, and permanently storing the data in the backtracking buffer area; triggering a fourth storage instruction when the total amplitude of the vibration signal or the amplitude of any characteristic frequency exceeds a preset safety threshold value, and permanently storing the data in the backtracking buffer area; When receiving an externally input manual trigger instruction, triggering a fifth storage instruction, and permanently storing the data in the backtracking buffer.
7. The multi-mode triggered turbine vibration data integrity storage method of claim 1, further comprising: Recording the time stamp of each finally confirmed permanent storage trigger event; calculating the occurrence frequency of the triggering event in a preset rolling time window to generate a triggering density index; And comparing the trigger density index with a group of trigger density thresholds which are preset and are associated with the equipment health state grades, and generating a grading index representing the equipment health state evolution trend.
8. A multi-mode triggered turbine vibration data integrity storage system comprising: The backtracking buffer area establishing module is used for establishing a backtracking buffer area for storing recent historical vibration data; The dynamic reference spectrum morphology vector calculation module is used for processing vibration signals acquired in real time into current spectrum morphology vectors and calculating dynamic reference spectrum morphology vectors representing recent stable states based on recent historical vibration data in the backtracking buffer area; the real-time monitoring loop module is used for calculating a first morphological similarity between the current spectrum morphological vector and the dynamic reference spectrum morphological vector in real time, and triggering a first storage instruction to permanently store data in the backtracking buffer area when the first morphological similarity is lower than a first preset threshold value; The second form similarity calculation module is used for pre-storing an generated frequency spectrum form vector representing the reference health state of the steam turbine; and the reference calibration loop module is used for monitoring drift caused by chronic degradation when the second form similarity is lower than a second preset threshold value, triggering a second storage instruction and permanently storing the data in the backtracking buffer.
9. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, performs the steps of a multi-mode triggered turbine vibration data integrity storage method as claimed in any one of claims 1 to 7.
10. A computer readable storage medium storing a computer program, wherein the computer program when executed by a processor implements the steps of a multi-mode triggered turbine vibration data integrity storage method of any one of claims 1-7.

Description

Multi-mode triggered turbine vibration data completeness storage method and related device Technical Field The invention belongs to the technical field of turbine fault monitoring, and relates to a multimode triggered turbine vibration data completeness storage method and a related device. Background In the current operation monitoring of large-scale high-speed rotating machinery such as a steam turbine, a commonly adopted technical scheme is to judge the running state of the machine set by monitoring the total amplitude of a vibration signal or the amplitude of a specific frequency and judging whether the energy related indexes exceed a preset threshold value or not, when the threshold value is triggered, the system starts the permanent storage of related vibration data, and the method can effectively respond and record the fault type with positive correlation between the fault severity and the vibration energy increase, however, during the actual running process of the machine set, the physical performance of part of critical faults in the initial stage of germination is not the increase of the total energy of vibration, but the change of the internal spectrum structure of a signal, such as new micro cracks or early slight rubs in the rotor, can cause new harmonic waves or side frequency components to appear in the signal, the energy of the new components is generally submerged in the background noise of the normal running of the machine set, and the storage mechanism based on the energy threshold value is not sufficiently triggered, so that the original data containing the fault origin information is discarded as a normal signal, thereby a rational monitoring omission is generated. In order to solve the problem, if the threshold value of energy triggering is directly reduced, another technical problem is caused in engineering application, namely harmless vibration caused by normal working condition fluctuation or load change of a machine set is caused, data storage is frequently triggered, a large amount of data without diagnostic value is generated, the situation not only occupies storage resources rapidly, but also causes interference to subsequent manual diagnosis analysis, so that the path does not have engineering practicability, therefore, the prior art has a technical dilemma in a triggering mechanism of data acquisition, 1, in order to ensure completeness of diagnosis analysis, a data acquisition system needs to have high sensitivity to weak signal structure change indicating early faults, 2, in order to ensure validity of data, the data acquisition system needs to have high tolerance to normal working condition fluctuation which does not represent real faults, 3, the prior technical method for judging depending on absolute magnitude of vibration energy is not met simultaneously due to single monitoring dimension, the technical requirements of the two mutual constraints are remained on response to vibration intensity, and the technical means for directly evaluating internal structural stability of vibration signals is lacked. The monitoring logic which only depends on the vibration intensity is not an example, even though some technical schemes which seem to optimize the triggering strategy are not provided, the kernel of the monitoring logic can not get rid of the dependence on the absolute magnitude of energy, so that the principle defects also exist, for example, the Chinese patent with the authority bulletin number of CN112362323B discloses a data storage method of a steam turbine generator unit vibration online monitoring and fault diagnosis system, although the method tries to improve the intelligent degree of data storage by setting various triggering conditions such as an alarm value, an amplitude change threshold value, an accumulated change quantity threshold value and the like, all judgment bases are instantaneous amplitude, change rate and accumulated increment, and the control logic is still a direct or indirect measure of the vibration energy, and is essentially micro-crack in the strength dimension, so that the method still can not effectively capture a key data chain of fault origin for forming a monitoring blind area due to the fact that the amplitude change is completely possible to be lower than all the threshold values. Therefore, how to build an early structural change of the dynamic characteristics of the unit by directly monitoring the stability of the spectrum morphology of the vibration signal without depending on the absolute magnitude of the vibration energy, so as to ensure the integrity capture of weak early failure information while avoiding invalid data recording is the technical problem to be solved by the invention. Disclosure of Invention The invention aims to provide a multimode triggering steam turbine vibration data completeness storage method and a related device, which are used for solving the technical problems that in the prior art, data storage triggering