Search

CN-121976920-A - Wind generating set blade root abnormality monitoring method and monitoring system

CN121976920ACN 121976920 ACN121976920 ACN 121976920ACN-121976920-A

Abstract

The invention discloses a method and a system for monitoring blade root abnormality of a blade of a wind generating set. The monitoring method comprises the following steps of obtaining a blade root torque related historical data file of a wind generating set, carrying out data modeling on the obtained blade root torque data file, obtaining the current health state of the blade root through real-time data and an algorithm model, and carrying out grading operation and maintenance according to abnormal risk grades if the blade root is judged to be in an abnormal state. The monitoring system is used for realizing the method and comprises a signal acquisition device, a data storage and management device, a data processing device, an abnormal grade judging device, a wind generating set main control device and a fault diagnosis module. The method can monitor the blade root state of the wind power blade in real time, on line, with high sensitivity and high reliability, and realize accurate identification and early warning of early damage of the blade root.

Inventors

  • ZHANG KUN
  • GUO ZIQIANG
  • ZENG YIMING
  • WANG WEIYU
  • JIANG ZHUOFU

Assignees

  • 东方电气风电股份有限公司

Dates

Publication Date
20260505
Application Date
20260121

Claims (10)

  1. 1. A method for monitoring abnormality of blade root of wind generating set is characterized by comprising the following steps: step 1, acquiring a blade root torque related historical data file of a wind generating set; Step 2, carrying out data modeling on the obtained blade root torque data file; Step 3, obtaining the current health state of the blade root through real-time data and an algorithm model; And step 4, if the blade root is judged to be in an abnormal state, carrying out hierarchical operation and maintenance according to the abnormal risk level.
  2. 2. The method of claim 1, wherein in step 1, the historical data file is a historical data set of all blade root torques and related operating parameters collected during a continuous operation time of the blade root of the wind turbine generator system after full inspection and no abnormality is confirmed.
  3. 3. The method for monitoring the blade root health state of the automobile according to claim 2, wherein in the step 2, data modeling is to construct a reference model of the blade root health state based on the historical data of the step 1 through data analysis and model storage of two core links; The data analysis comprises the steps of performing deep cleaning, characteristic engineering and mode learning on the obtained historical blade root torque data to generate a historical data analysis result capable of comprehensively representing the running characteristics of the blade root in a normal state; The model storage comprises the steps of structurally storing characteristic parameters, operation boundaries and mode rules extracted from the historical data analysis results in a database, so as to form a high-fidelity blade root torque normal state data model which can be called in real time.
  4. 4. The method of claim 3, wherein the data analysis performs comprehensive learning and analysis on the plurality of data dimensions obtained in step 1 by integrating a plurality of machine learning algorithms, and outputs a corresponding quantized data analysis result.
  5. 5. The method of claim 4, wherein the machine learning algorithm performs linear regression analysis for trend fitting, envelope analysis for extracting signal envelope features, correlation analysis for mining correlations between parameters, and random forest integration learning algorithm for processing high-dimensional nonlinear problems.
  6. 6. The method of claim 5, wherein the step 3 of obtaining the current health status of the blade root comprises two continuous processes of real-time data analysis and running status judgment; The real-time data analysis calls the established algorithm combination in the step 2, and the real-time calculation is carried out on the blade root torque data acquired in real time to obtain a real-time data analysis result which is the same in scale and standard as the historical data; And (2) the running state judgment is carried out according to the real-time data analysis result, and the running state judgment is dynamically compared and differentially evaluated with the blade root torque normal state data model stored in the step (2) through a rule abstraction method, so that whether the blade root is in an abnormal state currently is judged.
  7. 7. The monitoring method according to claim 6, wherein the rule abstraction method is an expert empirical method or a statistical abnormality judgment method; When the statistical abnormality judgment method is adopted, calculating the offset of the real-time data relative to the normal model, and judging whether the offset and the offset degree thereof are offset or not based on the quantification result; The statistical abnormality judgment method comprises 3 sigma criterion, isolated forest algorithm, Z-score standardization and box diagram method.
  8. 8. The monitoring method according to claim 7, wherein the step 4 comprises automatically triggering corresponding operation and maintenance processes according to the abnormal level determined by the rule abstraction method, and arranging on-site personnel to execute hierarchical operation and maintenance according to the determination result; the hierarchical operation and maintenance system automatically decides and recommends to adopt different operation and maintenance strategies according to the quantitative evaluation result of the abnormal degree of the blade root, wherein the strategies comprise immediate emergency treatment, planned shutdown treatment or enhanced monitoring delay treatment.
  9. 9. A wind generating set blade root abnormality monitoring system for the method of any one of claims 1 to 8, the system comprising a signal acquisition device, a data storage and management device, a data processing device, an abnormality level judgment device, a wind generating set main control device and a fault diagnosis module; The signal acquisition device is used for acquiring blade root torque related data files in real time at high frequency in the running process of the wind generating set; the data storage and management device is used for uniformly storing, classifying and managing the blade root torque related data file and the historical data file which are acquired in real time currently; The data processing device is internally provided with a machine learning algorithm and is used for carrying out data analysis on a currently acquired real-time data file and calculating a real-time state index representing the current health state of the blade root; The abnormal grade judging device is internally provided with an expert experience method or a statistical abnormal judging method and is used for diagnosing abnormal states, judging abnormal grades and then sending alarm or stop signals of different grades to the main control device according to judging results; the main control device of the wind generating set is used for sending blade root abnormality alarm signals of corresponding grades to a wind farm central control room after receiving the abnormality signals, and converting the emergency stop signals into a pitch-withdrawing stop instruction if receiving the emergency stop signals; the fault diagnosis module is used for acquiring a blade-collecting and stopping instruction and executing blade-collecting and stopping actions, so that the wind generating set is finally protected.
  10. 10. The monitoring system according to claim 9, wherein the data processing device is provided with an algorithm library comprising machine learning algorithms for data modeling and real-time analysis, and the abnormality level determination device is provided with a rule library comprising expert experience methods or statistical abnormality determination methods for final diagnosis and decision.

Description

Wind generating set blade root abnormality monitoring method and monitoring system Technical Field The invention relates to a method and a system for monitoring abnormality of a blade root of a blade of a wind generating set, and belongs to the technical field of wind generating sets. Background With the deep promotion of 'carbon peak, carbon neutralization' strategic targets in China, the wind power industry is accelerated to lead to transformation to large-scale and high-efficiency technologies, and high-power wind turbines become the main stream of the market. Meanwhile, the wind power development range is continuously expanded to remote inland, high-altitude areas and deep sea areas, so that the wind power plant is in complex and severe working condition environments such as high cold, strong wind, congelation, high salt spray corrosion and the like for a long time. Such environments place extremely high structural strength and durability demands on wind blades, particularly on blade roots, which are critical components for carrying and transmitting all aerodynamic loads. On the other hand, under the driving of increasingly strong market competition, the wind power blade generally adopts a lightweight design to improve the power generation efficiency and reduce the cost. However, lightweight designs are often accompanied by reduced material usage and reduced structural rigidity, which somewhat weakens the load bearing safety margin at the root location. The blade root serves as a core component for connecting the blade and the hub, and is subjected to huge alternating fatigue loads, extreme gust impacts and complex composite stresses for a long time. Under the double-factor superposition effect of light weight and severe environment, the risks of cracks, bolt pretightening force failure, structural damage and even fracture at the blade root part of the blade are obviously increased. Once structural faults such as blade root fracture and the like occur, the blade is quite likely to fall off, even the catastrophic accident of the whole unit inverted tower is caused, and not only is huge direct economic loss caused, but also serious threat is formed to the safety of wind farm operation and maintenance personnel and peripheral facilities. At present, the monitoring means for wind power blades are mostly concentrated on the whole vibration, surface damage (such as unmanned aerial vehicle visual detection) or front edge corrosion of the blades, and an effective and economic online monitoring method for hiding early potential damage of blade root parts is lacking. The traditional periodic manual inspection mode has the defects of early warning lag, dependence on experience and incapability of grasping the structural health state in real time, and is difficult to discover the hidden defect of initial cracks in the blade root in time. Therefore, a technology and a method for real-time, online and accurate monitoring of the health condition of the blade root of the wind power blade are urgently needed in the field so as to early warn in time at the early stage of damage, thereby providing decision support for preventive operation and maintenance, effectively avoiding the occurrence of disastrous accidents and ensuring safe and stable operation of the wind power generation set. Disclosure of Invention The invention aims to solve the problems, and provides a method and a system for monitoring the blade root abnormality of a wind turbine generator system blade, which can monitor the state of the blade root of the wind turbine generator system blade in real time, on line, with high sensitivity and high reliability, and realize accurate identification and early warning of early damage of the blade root. The technical scheme adopted by the invention is as follows: A wind generating set blade root abnormality monitoring method comprises the following steps: step 1, acquiring a blade root torque related historical data file of a wind generating set; Step 2, carrying out data modeling on the obtained blade root torque data file; Step 3, obtaining the current health state of the blade root through real-time data and an algorithm model; And step 4, if the blade root is judged to be in an abnormal state, carrying out hierarchical operation and maintenance according to the abnormal risk level. And (3) generating a normal corresponding relation database by the historical data file in the step (1), providing a reference basis for the establishment of a subsequent data model, and ensuring that the model is established on the basis of an actual running state. The built model in the step 2 can deeply mine the internal deep relation between the torque data in the step 1 and the health state of the blade root, and convert the relation into a determinable standard, so as to find a new reference object for the health state of the blade root. And step 3, carrying the real-time data into the algorithm model in the step 2, and further ob