CN-122014536-A - Online monitoring, early warning and waving quantity prediction method for waving deformation of in-service wind power blade

CN122014536ACN 122014536 ACN122014536 ACN 122014536ACN-122014536-A

Abstract

The invention belongs to the technical field of wind power blade deformation monitoring, and in particular relates to an in-service wind power blade waving deformation online monitoring, early warning and waving amount prediction method, which comprises the steps of synchronously and continuously collecting static data of a blade through a satellite positioning device; the method comprises the steps of obtaining static representative points and static reference plane equations of blade tips of all blades, obtaining dynamic coordinate training sequences of all blades, constructing and calibrating an independent self-adaptive Kalman filtering model for each blade, starting the self-adaptive Kalman filtering model corresponding to each blade to obtain a filtered dynamic coordinate sequence, calculating to obtain waving deformation of the blade, judging the state of the blade based on a preset multi-stage early warning threshold value, executing a corresponding control strategy, predicting the waving deformation of the blade in a future period, and sending multi-stage early warning prompt according to a prediction result. The invention can acquire the waving deformation condition of the blade in real time in the running process, and feeds back to the wind motor group in real time, and adjusts the control strategy of the wind power blade in advance.

Inventors

LI CAILIN
GUO BAOYUN
LI NING
LV CHUNJIANG
ZENG YIMING
WANG ZHIYONG
ZHANG LEIAN

Assignees

山东理工大学

Dates

Publication Date: 20260512
Application Date: 20260403

Claims (10)

1. The method for on-line monitoring, early warning and waving quantity prediction of the waving deformation of the in-service wind power blade is characterized by comprising the following steps: S1, respectively fixing satellite positioning devices at blade tips of all blades of a wind generating set, synchronously and continuously collecting static data of the blades under the condition that the blades are kept static, monitoring a data resolving type in real time by utilizing a main control system of the wind generating set, and only reserving data with the resolving type being a fixed solution as effective static data; S2, acquiring static representative points of the blade tips of all the blades and a reference plane equation when the blades are static by using effective static data; S3, synchronously and continuously collecting dynamic deformation data of the blades and preprocessing the dynamic deformation data under typical operation conditions of different wind speeds and corresponding unit rotating speeds to form a dynamic coordinate training sequence of each blade; s4, constructing and calibrating an independent self-adaptive Kalman filtering model for each blade, calibrating an optimal noise covariance matrix initial value corresponding to different operation conditions for each blade by utilizing a dynamic coordinate training sequence corresponding to each blade through an optimization algorithm, and establishing a mapping relation library of unit rotating speed-noise covariance matrix initial values; S5, according to the current unit rotating speed, calling a corresponding optimal noise covariance matrix initial value from a mapping relation library to serve as an initial parameter, and starting an adaptive Kalman filtering model corresponding to each blade to obtain a filtered dynamic coordinate sequence; S6, bringing the filtered dynamic coordinate sequence into a reference plane equation, and calculating the vertical distance of the dynamic deformation data point deviating from the reference plane to obtain the waving deformation of the blade; S7, transmitting the waving deformation to a main control system in real time, judging the state of the blade based on a preset multi-stage early warning threshold value, and executing a corresponding control strategy; S8, acquiring real-time wind speed data and future wind speed prediction data, constructing a piecewise polynomial regression model, predicting the blade waving deformation in a period of time in the future, and sending out multi-stage early warning prompt according to a prediction result.
2. The method for online monitoring, early warning and predicting the waving deformation of the in-service wind power blade according to claim 1, wherein in the step S1, the satellite positioning device adopts an RTK positioning module or a PPK positioning module, the data acquisition frequency of the satellite positioning device is set to be 20 times per second, under the condition that a wind power generator unit is stopped and the blade is kept static, the satellite positioning devices are controlled by a main control system to synchronously and continuously acquire static data of the blade for 10 minutes, during the acquisition, the main control system monitors the data calculation type in real time, only the data with the calculation type of a fixed solution is reserved, and the data with the calculation type of a floating solution and a single-point solution is removed.
3. The online monitoring, early warning and waving amount predicting method for the waving deformation of the in-service wind power blade according to claim 1, wherein in the step S2, a least square sphere fitting algorithm is adopted to process collected static data, a space sphere model is simulated to represent the space distribution characteristics of the blade tip in a static state, the sphere center is extracted as a static representative point position of the blade tip, a reference plane equation of the blade when the blade is static is determined based on the static representative point position, and a plane constant is determined.
4. The method for online monitoring, early warning and predicting the waving deformation of an in-service wind power blade according to claim 1, wherein in the step S3, the acquisition time period under each typical working condition is not less than 30 minutes, and the method is characterized in that Screening the dynamic coordinate sequences in the dynamic deformation data by a criterion to obtain dynamic coordinate training sequences of blade tips of all blades for model training: ; In the formula, Representing a dynamic coordinate training sequence of the blade tip of a kth blade at the moment t; 、、 and respectively representing the X-axis, Y-axis and Z-axis coordinate values of the blade tip of the kth blade at the moment t.
5. The method for online monitoring, early warning and predicting the flapping deformation of an in-service wind power blade according to claim 4, wherein in S4, the method is based on each typical working condition Constructing and calibrating an independent self-adaptive Kalman filtering model for each blade; Setting the state vector of the kth blade at the moment t Expressed as: ; In the formula, 、、 The coordinate values of the X axis, the Y axis and the Z axis of the blade tip of the kth blade at the moment t under the global coordinate system are respectively represented and are the real position states to be estimated; 、、 Respectively representing the motion speeds of the blade tip of the kth blade in the coordinate directions of an X axis, a Y axis and a Z axis at T moment to be estimated; by maximum likelihood estimation or expectation maximization algorithm, the optimal noise covariance matrix initial value corresponding to different operation conditions is marked for the kth blade, including the optimal process noise covariance matrix Optimal observation noise covariance matrix And establishing a mapping relation library of the set rotating speed-noise covariance matrix initial value.
6. The method for online monitoring, early warning and predicting the flapping deformation of in-service wind power blades according to claim 5, wherein in S5, after the blades are put into real-time monitoring operation, the main control system synchronously receives the real-time coordinate sequences uploaded by the satellite positioning devices , wherein, Representing the actual coordinate sequence of the kth blade tip at time t, 、、 Respectively representing the actual coordinate values of the tip of the kth blade at t moment in the X axis, the Y axis and the Z axis under the global coordinate system; calling corresponding from the mapping relation library according to the current unit rotating speed 、 As initial parameters, the adaptive Kalman filtering model corresponding to each blade is started to For observation input, the adaptive fusion of the model and observation is realized by the iterative process of state prediction and observation update and dynamically adjusting the noise covariance parameter according to the real-time innovation sequence, and finally the optimal estimation of each blade state vector is output, the position component is extracted from the optimal estimation, and the filtered dynamic coordinate sequence is obtained 。
7. The method for online monitoring, early warning and predicting the waving deformation of in-service wind power blades according to claim 6, wherein in S6, the waving deformation of the kth blade at the moment t The calculation formula of (2) is as follows: ; In the formula, 、、 The components of the plane normal vector in the directions of the X axis, the Y axis and the Z axis of the global coordinate system are respectively shown, and D is a plane constant.
8. The method for online monitoring, early warning and predicting the waving deformation of an in-service wind power blade according to claim 1, wherein in S7, based on the elastic deformation limit and the historical operation data of the blade material, three-level early warning thresholds are preset, and are respectively: ; In the formula, 、、 Respectively representing the first-level, second-level and third-level thresholds of the kth blade; The peak value average value of the waving deformation of the kth blade under the normal working condition; The standard deviation of the waving deformation of the kth blade; the judging of the blade state and executing the corresponding control strategy are specifically as follows: When (when) When the wind turbine generator system is started, the primary early warning is triggered, wherein the main control system outputs prompt information to a monitoring center of the wind turbine generator system to remind operation and maintenance personnel of paying attention to the state of the blades; When (when) Triggering a secondary early warning, namely adjusting the pitch angle by a main control system and reducing the pneumatic load of the blade; When (when) And triggering three-stage early warning, namely immediately starting an emergency stop program, feathering the blades to 90 degrees and activating mechanical brake.
9. The method for online monitoring, early warning and predicting the flapping deformation of an in-service wind power blade according to claim 8, wherein in S8, real-time wind speed data is obtained by using a wind field velocimeter And acquire future 24-hour wind speed prediction data provided by the wind field For a pair of 、 Respectively carrying out max-min standardization processing to obtain standardized wind speed data And wind speed prediction data ; Dividing four wind speed sections, namely a low wind speed section, a medium wind speed section, a high wind speed section and an extreme wind speed section according to the wind speed; For each wind speed interval, use is made of Constructing a third-order polynomial regression prediction model of a kth blade: ; In the formula, A waving deformation prediction sequence of the kth blade is represented; 、、、 The model coefficient of the mth wind speed interval is obtained by fitting a historical wind speed-waving deformation matching data set through a weighted least square method; Will be A third-order polynomial regression prediction model with the kth blade is carried into to obtain a future waving deformation prediction sequence of the kth blade Will be The actual waving deformation amount prejudgement sequence is restored by inverse standardization Based on And sending out a multi-stage early warning prompt by combining with a preset multi-stage early warning threshold.
10. The method for online monitoring, early warning and predicting the flapping deformation of an in-service wind power blade according to claim 9, wherein in S8, for any one specific future time t, the method is based on And 、、 The method sends out multi-stage early warning prompts, specifically: When (when) When the method is used, a primary early warning prompt is triggered, namely a main control system outputs early warning prompt information to a monitoring center to inform operation and maintenance personnel of the rising trend of future waving deformation and remind the operation state of paying attention in advance; When (when) When the system is used, a secondary early warning prompt is triggered, namely the main control system is combined Actively adjusting the pitch angle, and intervening in advance to reduce the expected pneumatic load; When (when) Triggering three-level early warning prompt: linkage of main control system And starting an emergency stop preparation program, and simultaneously immediately informing operation and maintenance personnel to be in place to finish emergency response preparation work.

Description

Online monitoring, early warning and waving quantity prediction method for waving deformation of in-service wind power blade Technical Field The invention belongs to the technical field of wind power blade deformation monitoring, and particularly relates to an in-service wind power blade waving deformation online monitoring, early warning and waving amount prediction method. Background Wind power generation has been widely applied and rapidly developed at present because of the characteristics of abundant resources, environmental friendliness, mature technology and the like, and has become a third main power source after thermal power and hydropower. The wind generating set is core equipment for realizing wind energy conversion, wherein blades are used as key components for capturing wind energy and converting the wind energy into mechanical energy, and the structural safety and the running state of the wind generating set are directly related to the performance of the whole machine, the power generation efficiency and the service life. Along with the development of wind power technology to large-scale and high-power directions, the length of blades is continuously increased, the structure is increasingly complex, the actions of aerodynamic load, gravity load, inertial load, environmental corrosion and the like born by the wind power technology in the running process are also increasingly severe, structural fatigue, deformation and even fracture are extremely easy to cause, and the safety and economic running of a unit are seriously affected. Flapwise deformation of a blade, i.e. bending vibrations in a direction perpendicular to the plane of rotation, is a common form of dynamic response of a blade in complex wind conditions. Excessive waving deformation not only can lead to reduced aerodynamic performance and reduced power generation efficiency, but also can cause serious consequences such as collision of blades and towers, structural fatigue accumulation, damage of key parts and the like, and even cause catastrophic accidents. In recent years, events such as blade breakage, machine set shutdown and the like caused by overlarge blade waving occur at home and abroad, and serious economic loss and potential safety hazard are caused. At present, the state monitoring of in-service blades in the wind power industry mainly depends on periodic inspection, off-line detection and laboratory simulation test, and lacks real-time, on-line and continuous monitoring means for blade waving deformation. Although the traditional strain gage, acceleration sensor and other modes can be used for local deformation measurement, the whole dynamic deformation state of the blade is difficult to comprehensively reflect, and the strain gage is easy to be interfered by environment, complex to install and poor in durability. In addition, the existing method focuses on fault diagnosis and health assessment after the accident, cannot realize real-time feedback and active control in the operation process, and is difficult to early warn and take adjustment measures in time at the initial stage of deformation occurrence, so that the operation load of the blade cannot be effectively reduced, fatigue damage is delayed, and the service life of a unit is prolonged. Therefore, the technology capable of monitoring the blade waving deformation in real time and carrying out early warning and trend prediction is researched, closed-loop linkage with a unit control system is realized, and the method has important significance for improving the running safety, reliability and economy of the wind turbine. Disclosure of Invention According to the defects in the prior art, the invention aims to provide the in-service wind power blade waving deformation on-line monitoring, early warning and waving quantity prediction method, which can acquire waving deformation conditions of wind power blades in the running process in real time, pre-judge subsequent deformation trend by combining future wind field prediction data, feed back to a wind motor set in real time, adjust control strategies such as steering, pitch angle and the like of the wind power blades in advance, provide data support for preventive maintenance of the blades, and develop targeted maintenance in time. In order to achieve the purpose, the invention provides an online monitoring, early warning and waving amount prediction method for waving deformation of an in-service wind power blade, which comprises the following steps: S1, respectively fixing satellite positioning devices at blade tips of all blades of a wind generating set, synchronously and continuously collecting static data of the blades under the condition that the blades are kept static, monitoring a data resolving type in real time by utilizing a main control system of the wind generating set, and only reserving data with the resolving type being a fixed solution as effective static data; S2, acquiring static representative points of