Wind driven generator and tower health state assessment method thereof
By obtaining the detection data of the tower and building a relationship model and threshold calculation model, the accuracy of the tower health status assessment of the wind turbine tower is solved, and intelligent monitoring of the tower health status is realized to reduce the risk of fan downward.
Patent Information
- Application Number
- CN202411005167.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-25
- Publication Date
- 2024-11-19
AI Technical Summary
The prior art is difficult to accurately and intelligently monitor and analyze the health status of wind turbine towers, which leads to difficulty in detecting fatigue damage in a timely manner, which may lead to the risk of fan downturning.
By acquiring the detection data of the tower, including the tower top displacement data, bending moment data at the connection of adjacent tower segments and unit SCADA data, a relationship model and threshold calculation model are constructed, and the health status of the tower is evaluated using adaptive thresholds to achieve Accurate and intelligent assessment of the health status of the tower.
This method can more accurately evaluate the health status of the tower, reduce the risk of fan turning over, and improve the safety and reliability of wind turbines.
Smart Images

Figure CN118979852A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of wind turbines, and in particular to a wind turbine and a tower health assessment method thereof. Background Art
[0002] With the development of large-scale wind turbines, the maximum tower height has reached 185 meters. In addition to steel towers, different tower structures such as hybrid towers and truss towers have emerged. During the operation of wind turbines, the tower bears the gravity and wind loads of blades and nacelles. The bending moments include compression, tension, twisting, bending, etc., and the direction and size change all the time. Fatigue damage is prone to occur. If it is not discovered and reinforced in time, it may cause the wind turbine tower to collapse in serious cases, causing serious losses.
[0003] At present, the monitoring of the tower is mainly for the weld monitoring, vibration acceleration monitoring, and inclination monitoring of the flange connection of each section of the tower. The health status and sway displacement of the tower structure are judged by analyzing the relevant displacement, vibration acceleration, and inclination signals. In general, the current online monitoring of the wind turbine tower is mainly for inclination monitoring and analysis, and there is less analysis of vibration and strain monitoring signals. In addition, the analysis of inclination signals is mostly time domain amplitude and frequency domain resonance analysis. The early warning is often based on artificially set thresholds, making it difficult to accurately and intelligently monitor and analyze the health status of the tower. Summary of the invention
[0004] The purpose of the present invention is to provide a method for evaluating the health of a wind turbine tower, which can improve the technical problem existing in the prior art that it is difficult to accurately and intelligently monitor and analyze the health status of the tower.
[0005] Another object of the present invention is to provide a wind turbine generator that can improve the technical problem in the prior art that it is difficult to accurately and intelligently monitor and analyze the health status of the tower.
[0006] The embodiments of the present invention can be implemented in the following ways:
[0007] A wind turbine tower health assessment method, the wind turbine tower health assessment method is used to assess the health status of a wind turbine tower; the tower comprises at least two tower sections connected in sequence; the wind turbine tower health assessment method comprises:
[0008] Acquire detection data of the tower; wherein the detection data includes tower top displacement data, bending moment data at the connection between adjacent tower sections, and unit SCADA data;
[0009] The relational model outputs theoretical data corresponding to the detection data according to the detection data;
[0010] The threshold calculation model calculates the difference and the adaptive threshold according to the detection data and the theoretical data; wherein the threshold calculation module comprises an extreme value data set, a window data set, an initial threshold and an initial deviation value; the initial threshold and the initial deviation data, the initial threshold is obtained according to the extreme value data set, and the initial deviation data is obtained according to the window data set; the extreme value data set is used to calculate the adaptive threshold;
[0011] The calculated difference is compared with the adaptive threshold, and the health status of the wind turbine tower is evaluated according to the difference between the calculated difference and the adaptive threshold.
[0012] Optionally, the wind turbine tower health status assessment method further includes a step of constructing a relationship model; the step of constructing the relationship model includes:
[0013] Constructing a modeling data set; wherein the modeling data set includes historical data of the tower in a healthy state, and the historical data includes historical tower top displacement data, historical bending moment data at the connection between adjacent tower sections, and historical unit SCADA data;
[0014] A relational model is obtained using the modeling data set and based on a DAE algorithm; wherein the relational model is used to output theoretical data corresponding to the input data.
[0015] Optionally, in the process of constructing the relational model, the data in the modeling data set is normalized, and the relational model is obtained based on the DAE algorithm using the normalized values.
[0016] Optionally, the wind turbine tower health status assessment method further includes the step of constructing a threshold value calculation model; the step of constructing a threshold value calculation model includes:
[0017] Calculate historical deviation data based on the modeling data set and modeling theoretical data output by the relational model and corresponding to the modeling data set;
[0018] An extreme value data set and a window data set are constructed according to the historical deviation data, and the initial threshold is calculated according to the extreme value data set, and the initial deviation data is calculated according to the window data.
[0019] Optionally, the step of calculating historical deviation data according to the modeling data set and the modeling theoretical data output by the relational model and corresponding to the modeling data set comprises:
[0020] Obtaining a historical difference value according to the historical bending moment data in the modeling data set and the modeling theoretical bending moment data output by the relationship model and corresponding to the historical bending moment data;
[0021] converting the historical difference into historical deviation data;
[0022] The steps of constructing an extreme value data set according to the historical deviation data include:
[0023] Arrange the plurality of historical deviation data from large to small, and use the value of the t%th data after arrangement as the extreme value threshold value;
[0024] The data in the historical deviation data that is greater than or equal to the extreme value threshold value is constructed into extreme value data.
[0025] Optionally, the formula for converting the historical difference into historical deviation data is:
[0026] x i =ΔF i -M i ,i=d+1,...,N
[0027]
[0028] Among them, x i is the historical deviation data at the i-th time point; M i is the historical deviation average of the d time series before the i-th time point; d is a preset positive integer; N is the number of time points in the modeling data set; ΔF i is the historical difference at the i-th time point.
[0029] Optionally, the step of calculating the initial threshold value according to the extreme value data set includes:
[0030] x i ′=x i -T
[0031]
[0032] Z q,i =M i +T+b
[0033]
[0034] Among them, x′ i is the extreme value data at the i-th time point, x i is the historical deviation data at the i-th time point; T is the extreme value threshold; q is the preset extreme value probability; Z q,i is the threshold value.
[0035] Optionally, there are multiple historical bending moment data, and the modeling theoretical data includes multiple modeling theoretical bending moment data, and the multiple modeling theoretical bending moment data correspond one-to-one to the multiple historical bending moment data;
[0036] The step of calculating historical deviation data according to the modeling data set and the modeling theoretical data corresponding to the modeling data set output by the relational model comprises:
[0037] The difference between the modeling theoretical bending moment data and the corresponding historical bending moment data is calculated, and the largest one of the multiple differences is used as the historical deviation data.
[0038] A wind turbine generator comprises a tower, a strain sensor, a displacement detection member, a monitoring and control system, and a health status assessment system. The tower comprises at least two tower sections. The strain sensor is arranged at the connection between two adjacent tower sections to obtain the bending moment data at the connection between the adjacent tower sections. The displacement detection member is used to obtain the displacement data of the tower top. The strain sensor, the displacement detection member, and the monitoring and control system are electrically connected to the health status assessment system. The monitoring and control system is used to transmit the unit SCADA data to the health status assessment system. The health status assessment system is used to execute the above-mentioned wind turbine tower health status assessment method to perform a health assessment on the tower.
[0039] Optionally, the lowest one of the at least two tower sections is a bottom tower section, and the highest one of the at least two tower sections is a top tower section; the displacement detection member comprises a first inclination sensor and a second inclination sensor, the first inclination sensor being arranged on the bottom tower section to obtain the inclination angle of the bottom of the tower; the second inclination sensor being arranged on the top tower section to obtain the inclination angle of the top of the tower;
[0040] The tower top displacement data is calculated according to the following formula:
[0041]
[0042] Among them, d l is the displacement data value of the tower top; l is the setting direction of the first inclination sensor and the second inclination sensor; H is the tower height; θ l,塔顶 is the inclination angle of the tower top; θ l,塔底 It is the inclination angle of the tower bottom in direction l.
[0043] The beneficial effects of the wind turbine and tower health status assessment method provided by the embodiment of the present invention include:
[0044] The wind turbine tower health status assessment method provided by the embodiment of the present invention includes obtaining detection data of the tower; wherein the detection data includes tower top displacement data, bending moment data at the connection of adjacent tower sections and unit SCADA data; the relationship model outputs theoretical data corresponding to the detection data according to the detection data; the threshold calculation model calculates the calculated difference and the adaptive threshold according to the detection data and the theoretical data; wherein the threshold calculation module has an extreme value data set, a window data set, an initial threshold and an initial deviation value; the initial threshold and the initial deviation data, the initial threshold is obtained according to the extreme value data set, and the initial deviation data is obtained according to the window data set; the extreme value data set is used to calculate the adaptive threshold; the calculated difference and the adaptive threshold are compared, and the healthy state of the wind turbine tower is assessed according to the difference between the calculated difference and the adaptive threshold. The assessment method takes multiple detection data as input, so that the healthy state of the tower can be assessed more accurately and intelligently.
[0045] An embodiment of the present invention further provides a wind turbine, which includes a health status assessment system for executing the above-mentioned wind turbine tower health status assessment method, and thus also has the beneficial effect of being able to more accurately and intelligently assess the health status of the tower. BRIEF DESCRIPTION OF THE DRAWINGS
[0046] The above features and advantages of the present invention can be better understood after reading the detailed description of the embodiments of the present disclosure in conjunction with the following drawings. In the drawings, the components are not necessarily drawn to scale, and components with similar related properties or features may have the same or similar reference numerals.
[0047] Figure 1 A schematic structural diagram of a wind turbine generator according to one aspect of the present invention is shown;
[0048] Figure 2 A step diagram of a method for evaluating the health status of a wind turbine tower provided according to one aspect of the present invention is shown;
[0049] Figure 3 A relationship diagram showing first bending moment data provided according to one aspect of the present invention and theoretical bending moment data outputted using a relationship model is shown;
[0050] Figure 4 A relationship diagram showing fourth bending moment data provided according to one aspect of the present invention and theoretical bending moment data outputted using a relationship model is shown;
[0051] Figure 5 A schematic diagram showing a comparison between a calculated difference and an adaptive threshold provided according to an aspect of the present invention is shown.
[0052] Reference numerals:
[0053] 100-wind turbine; 110-tower; 111-bottom tower section; 112-first intermediate tower section; 113-second intermediate tower section; 114-third intermediate tower section; 115-top tower section; 116-first connection; 117-second connection; 118-third connection; 119-fourth connection; 120-strain sensor; 131-first inclination sensor; 132-second inclination sensor. DETAILED DESCRIPTION
[0054] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. Note that the aspects described below in conjunction with the accompanying drawings and specific embodiments are only exemplary and should not be construed as limiting the scope of protection of the present invention in any way.
[0055] In the description of the present invention, it should be noted that if the terms "upper", "lower", "inside", "outside", "vertical" and the like appear, the orientation or position relationship indicated is based on the orientation or position relationship shown in the drawings, or is the orientation or position relationship in which the product of the invention is usually placed when used, rather than indicating or implying that the device or element referred to must have a specific orientation or be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention.
[0056] At the same time, it should be noted that the terms "first", "second", etc. are only used for distinguishing descriptions and cannot be understood as indicating or implying relative importance.
[0057] In the description of the present invention, it is also necessary to explain that, unless otherwise clearly specified or limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, an integral connection, or a detachable connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium, or the internal communication of two components, etc. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.
[0058] Figure 1 The schematic diagram of the structure of the wind turbine 100 provided in this embodiment is as follows: Figure 2 This is a step diagram of the wind turbine tower health status assessment method provided in this embodiment. Please refer to Figure 1 and Figure 2The embodiment of the present invention provides a wind turbine 100, and also provides a method for evaluating the health status of a tower 110 based on the wind turbine 100. Specifically, the wind turbine 100 includes a tower 110, a strain sensor 120 (also referred to as a load sensor), a displacement detection component, a monitoring and control system (not shown), and a health status evaluation system (not shown). The tower 110 includes at least two tower sections, and the strain sensor 120 is arranged at the connection between two adjacent tower sections. The bending moment data at the connection between adjacent tower sections is obtained through the strain sensor 120. The displacement detection component is used to obtain the displacement data of the tower top. The monitoring and control system (SCADA) is a system provided by the existing wind turbine 100, which is used to monitor the operation of the wind turbine 100 and obtain the SCADA data of the unit. Specifically, the SCADA data of the unit includes wind speed, power, wind angle, and pitch angle.
[0059] Specifically, in the present embodiment, the tower 110 includes five tower sections stacked up and down, and the five sleeve sections are respectively a bottom tower section 111, a first intermediate tower section 112, a second intermediate tower section 113, a third intermediate tower section 114 and a top tower section 115. Accordingly, the “adjacent tower section connections” referred to in the present embodiment include a first connection 116, a second connection 117, a third connection 118 and a fourth connection 119. The first connection 116 is the connection between the bottom tower section 111 and the first intermediate tower section 112, the second connection 117 is the connection between the first intermediate tower section 112 and the second intermediate tower section 113, the third connection 118 is the connection between the second intermediate tower section 113 and the third intermediate tower section 114, and the fourth connection 119 is the connection between the third intermediate tower section 114 and the top tower section 115. Furthermore, in this embodiment, the wind turbine 100 includes a plurality of strain sensors 120, and the plurality of strain sensors 120 are respectively used to obtain the bending moment data of the first connection 116, the second connection 117, the third connection 118, and the fourth connection 119, the bending moment data of the first connection 116 is the first bending moment data, the bending moment data of the second connection 117 is the second bending moment data, the bending moment data of the third connection 118 is the third bending moment data, and the bending moment data of the fourth connection 119 is the fourth bending moment data. It can be understood that in some other embodiments, the specific number of tower sections can also be set according to requirements, and accordingly, the number of obtained bending moment data varies accordingly.
[0060] Optionally, a plurality of strain sensors 120 are provided at each connection, and the plurality of strain sensors 120 are distributed along the circumferential direction of the tower 110. Optionally, in this embodiment, eight strain sensors 120 are provided at each connection, and it is understandable that in some other embodiments, the number of strain sensors 120 at each connection may also be set according to requirements.
[0061] In this embodiment, the displacement detection member includes a first inclination sensor 131 and a second inclination sensor 132. The first inclination sensor 131 is arranged on the bottom tower section 111, so as to obtain the inclination angle of the bottom of the tower 110. The second inclination sensor 132 is arranged on the top tower section 115, so as to obtain the inclination angle of the top of the tower 110. The tower top displacement data can be calculated using the detection data of the first inclination sensor 131 and the second inclination sensor 132. Specifically, the first inclination sensor 131 and the second inclination sensor 132 are both dual-axis inclination sensors.
[0062] The tower top displacement data can be calculated according to the following formula:
[0063]
[0064] Among them, d l is the displacement data value of the tower top; l is the setting direction of the first inclination sensor 131 and the second inclination sensor 132; H is the height of the tower 110; θ l,塔顶 is the inclination angle of the top of the tower 110; θ l,塔底 is the inclination angle of the bottom of the tower 110 in direction l.
[0065] The strain sensor 120, the displacement detection element and the monitoring and control system are all electrically connected to the health status assessment system, so that the health status assessment system obtains the bending moment data at the connection between adjacent tower sections from the strain sensor 120, the tower top displacement data from the displacement detection element, and the unit SCADA data from the monitoring and control system. The health assessment system uses the obtained bending moment data at the connection between adjacent tower sections, the tower top displacement data and the unit SCADA data to perform a health status assessment on the wind turbine tower.
[0066] Specifically, Figure 2 As shown, this embodiment also provides a wind turbine tower health status assessment method, and the health assessment system implements the health status assessment of the wind turbine tower by executing the wind turbine tower health status assessment method. The wind turbine tower health status assessment method provided by this embodiment is further described below:
[0067] The wind turbine tower health status assessment method specifically includes the following steps:
[0068] S01: Acquire detection data of the tower 110 .
[0069] The detection data set includes tower top displacement data, bending moment data at the connection between adjacent tower sections and unit SCADA data. In this embodiment, the bending moment data includes first bending moment data, second bending moment data, third bending moment data and fourth bending moment data.
[0070] Furthermore, considering the differences in sampling frequencies of the unit SCADA data, the tilt sensor and the strain sensor 120, the signals of each measuring point are averaged according to a preset time length, thereby obtaining the tower top displacement data, bending moment data and unit SCADA data at multiple time points. Optionally, the preset time length can be set to 1 minute.
[0071] S02: The relational model outputs theoretical data corresponding to the detection data based on the detection data.
[0072] The input neurons of the relational model take the detection data as input, and the output neurons of the relational model take the theoretical data corresponding to the detection data as output. Specifically, the input of the relational model is the tower top displacement data, the bending moment data and the unit SCADA data, and correspondingly, the output of the relational model is the theoretical tower top displacement data, the theoretical bending moment data and the theoretical unit SCADA data. In this embodiment, since the bending moment data includes the first bending moment data, the second bending moment data, the third bending moment data and the fourth bending moment data, correspondingly, the theoretical bending moment data output by the relational model includes the first theoretical bending moment data, the second theoretical bending moment data, the third theoretical bending moment data and the fourth theoretical bending moment data.
[0073] Furthermore, before executing step S02, the wind turbine tower health status assessment method further includes the step of constructing a relationship model. The step of constructing a relationship model includes:
[0074] S21: Build modeling dataset.
[0075] A modeling data set is constructed based on historical data of the tower 110 in a healthy state, where the historical data includes historical tower top displacement data, historical bending moment data at the connection between adjacent tower sections, and historical unit SCADA data.
[0076] S22: Obtain a relational model using the modeling data set based on the DAE algorithm.
[0077] Using the autoencoder algorithm (DAE) in unsupervised learning, the unit SCADA data (wind speed, power, wind angle, pitch angle), bending moment data of each floor, and tower top displacement data are used as input and output neurons to establish a relationship model.
[0078] Furthermore, in the calculation, the data in the modeling data set needs to be normalized so that the value range of each parameter is between [-1, 1] and then substituted into the machine learning. Figure 3 The relationship diagram of the first bending moment data and the theoretical bending moment data outputted by the relationship model is shown. Figure 3 It can be seen that the two are basically linearly related. Figure 4 A relationship diagram of the fourth bending moment data and theoretical bending moment data outputted using the relationship model is shown.
[0079] S03: The threshold calculation model calculates the difference and the adaptive threshold according to the detection data and the theoretical data.
[0080] Since the local deviation and the adaptive threshold are calculated by a threshold calculation model, before executing step S03, the wind turbine tower health status assessment method may further include the step of constructing a threshold calculation model. Specifically, the step of constructing a threshold calculation model includes:
[0081] S31: Calculate historical deviation data based on the modeling data set and the modeling theoretical data corresponding to the modeling data set of the relational model data.
[0082] The modeling data set has multiple historical bending moment data at the same time point, and correspondingly, the modeling theoretical data output by the relational model has multiple modeling theoretical bending moment data at the same time point.
[0083] Therefore, when calculating the historical deviation data, the difference between the modeled theoretical bending moment data and the corresponding historical bending moment data can be calculated, and the largest one among the multiple differences can be used as the historical difference.
[0084] Specifically, the historical difference can be obtained according to the following method:
[0085] ΔF i =max(F i,m -Theory F i,m )
[0086] Where, ΔF i is the calendar difference value at the i-th time point; F i,m is the historical bending moment data at the mth position at the i-th time point; the theoretical F i,m is the modeling theoretical bending moment data at the mth point at the ith time point; m is a positive integer from 1 to n, and n is the number of tower segment connections. In the present embodiment, the tower 110 has five tower segments and four connections. Accordingly, in the present embodiment, n is 4.
[0087] The historical deviation data can be obtained based on the historical difference. The process of obtaining the historical deviation data specifically includes:
[0088] Based on the modeling data set, the historical difference value at each time point i obtained in step S31 is converted into historical deviation data in sequence. Specifically, the conversion is performed according to the following formula:
[0089] x i =ΔF i -M i ,i=d+1,...,N
[0090]
[0091] Among them, x i is the historical deviation data at the i-th time point; M i is the historical deviation average of the d time series before the i-th time point; d is a preset positive integer; N is the number of time points in the modeling data set; ΔF i is the historical difference at the i-th time point.
[0092] Since the number of time points in the modeling data set is N, the number of historical deviation data calculated in sequence for the data in the modeling data set is (Nd). In other words, the first d data in the modeling data set have no corresponding historical deviation data.
[0093] S32: Construct an extreme value dataset and a window dataset based on historical deviation data.
[0094] The steps to construct an extreme value dataset based on historical deviation data include:
[0095] S321: Arrange multiple historical deviation data from large to small, and use the value of the t%th data after arrangement as the extreme value threshold value
[0096] Arrange the (Nd) historical deviation data obtained in step S31 in descending order, and use the value T of the t% data after arrangement as the extreme value threshold. If (Nd)t% is not an integer, it can be rounded up by the method of rounding up to 1. The specific value of t can be selected based on experience, for example, t% can be set to 5%, 6%, 7%, etc.
[0097] S322: Construct an extreme value data set from the data in the historical deviation data that is greater than or equal to the extreme value threshold. i , and subtract the extreme value threshold T from the acquired data to form the extreme value data x′ i :
[0098] x′ i =x i -T.
[0099] Since the data used to construct the extreme value data set is the data in the local deviation that is greater than or equal to the extreme value threshold, the extreme value data x′ formed i The minimum value of is 0, which is convenient for extreme value data fitting. It should be noted that in this embodiment, in order to facilitate extreme value data fitting, the data that meets the conditions in the local deviation is converted into extreme value data x′ i It is understandable that in some other embodiments, the data satisfying the conditions in the local deviation can be directly used to form an extreme value data set according to the needs.
[0100] The steps to construct a window dataset based on historical deviation data include:
[0101] The historical deviation data of the last d time points in the modeling data set are combined into a window data set. Specifically, the specific value of d can be set by selecting a positive integer according to the requirements.
[0102] S323: Obtain an initial threshold value according to the extreme value data set.
[0103] According to extreme value theory, x′ i The distribution of is independent of the data set itself and has a compound exponential distribution, and its probability distribution density function is:
[0104]
[0105] Among them, γ is the parameter of the exponential distribution function.
[0106] Since x′ i The number and value of have been determined, so the value of the shape parameter γ can be obtained by fitting the above probability distribution density function.
[0107] Threshold Z q,i It can be calculated according to the following formula:
[0108] Z q,i =M i +T+b;
[0109]
[0110] Among them, x′ i is the extreme value data; T is the extreme value threshold, q% is the preset extreme value probability, Z q,i is the threshold value.
[0111] The specific value of q% can be preset based on experience, etc. When a probability q corresponding to a threshold is manually specified, the "extreme value" b in the extreme value data set can be obtained. Optionally, the specific value of q% can be set to 1%, 2%, etc.
[0112] In the process of executing step S32, the obtained extreme value data set can also be regarded as the initial extreme value data set of the threshold value calculation model. At the same time, the obtained window data set is the initial window data set of the threshold value calculation model. The value of the shape parameter γ obtained by fitting the data in the initial extreme value data set and the historical deviation average value M obtained by fitting the data in the initial window data set are i The calculated threshold Z q,i is the initial threshold.
[0113] After the threshold calculation model is constructed, the threshold calculation model can be used to calculate the difference and the adaptive threshold according to the detection data obtained during the actual operation of the wind turbine 100 and the theoretical data obtained by inputting the detection data into the relationship model. The calculation steps include:
[0114] S331: Calculate the difference based on the detection data and the theoretical data obtained by inputting the detection data into the relationship model.
[0115] According to the formula for calculating historical differences in step S31, the one with the largest deviation value among the multiple connections at the current moment is calculated as the calculated difference value ΔF at the current moment i. i .
[0116] S332: Convert the calculated difference into a local deviation.
[0117] According to the step of calculating the historical deviation data in step S31, the difference ΔF is calculated. i Transformed into local deviation x i .
[0118] Specifically, when the current moment i is the first moment detected in real time, M i That is, the average value obtained by using the data in the initial window data set. As time moves backward, the data in the window data set is continuously updated, and the M calculated is i Constantly updated.
[0119] S333: The initial threshold is used as an adaptive threshold of the local deviation calculated at the first moment of real-time detection, and then the adaptive threshold and the local difference are compared, and the extreme value data set is updated according to the comparison structure.
[0120] Specifically, if the local deviation x i If the local deviation x is less than the extreme value threshold T, the extreme value data set will not be updated; i Greater than or equal to the extreme threshold T, and less than or equal to the threshold Z q,i-1 , then the local deviation x i The extreme value data x obtained by subtracting the extreme value threshold value T i′ is added to the extreme value data set, and the new shape parameter γ is obtained by refitting according to the steps in step S323, and the threshold Z is calculated using the new shape parameter q,i , and the recalculated threshold Z q,i As the adaptive threshold at the next time point; if the local deviation x i Greater than the threshold Z q,i-1 , an alarm is required.
[0121] S04: Compare the calculated difference with the adaptive threshold, and evaluate the health status of the wind turbine tower according to the difference between the calculated difference and the adaptive threshold.
[0122] The health status can be divided into three levels. Specifically, when the calculated difference is greater than the adaptive threshold, it means that the health status is poor; when the calculated difference is less than or equal to the adaptive threshold, and the deviation is greater than or equal to 0.1, it means that the health status is good; when the calculated difference is less than the adaptive threshold, and the deviation is less than 0.1, it means that the health status is excellent.
[0123] Figure 5 FIG. 4 is a schematic diagram showing a comparison between the calculated difference and the adaptive threshold provided in this embodiment. Specifically, Figure 5 The data in the figure are the 110 bending moment, inclination angle, and unit SCADA monitoring data of a 90-meter-long tower of a 2.2MW unit, which are obtained by testing using the wind turbine tower health status assessment method provided by the present invention. At the same time, when constructing the threshold meter model, t=5, that is, the value of the 5% maximum deviation between the theoretical bending moment and the actual bending moment in the modeling data is taken as the extreme threshold value T, q is set to 1%, and the threshold value of each time point is calculated in turn with 6 data points as the sliding window. The test results are as follows Figure 5 As shown, during the 4-hour operation, the calculated difference in the bending moment data of the tower 110 did not exceed the adaptive threshold and was in a good and excellent state.
[0124] The health status assessment method of the wind turbine 100 and its tower 110 provided by the embodiment of the present invention takes the bending moment signal at the connection of each layer as the target value, and simultaneously takes the tower top displacement and the unit SCADA data related measurement point data as input, performs unsupervised machine learning modeling based on the historical health status data, and obtains the relationship model between the target value and the input value; based on the health status data, calculates the maximum value of the deviation between the bending moment of each section of the tower 110 and the theoretical value of the model at the same time step; and then uses the adaptive threshold algorithm to analyze the starting threshold of the maximum deviation, thereby constructing a threshold calculation model. After substituting the actual operation data into the relationship model and the threshold calculation model, the health status of the tower 110 is judged based on whether the calculated difference exceeds the adaptive threshold, thereby achieving an accurate and intelligent assessment of the health status of the tower 110.
[0125] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by any technician familiar with the technology in the field within the technical scope disclosed by the present invention should be covered within the protection scope of the present invention.
Claims
1. A method for evaluating the health status of a wind turbine tower, wherein the method is used to evaluate the health status of a wind turbine tower; wherein: The tower includes at least two tower sections connected in sequence; the wind turbine tower health status assessment method includes: Acquire detection data of the tower; wherein the detection data includes tower top displacement data, bending moment data at the connection between adjacent tower sections, and unit SCADA data; The relational model outputs theoretical data corresponding to the detection data according to the detection data; The threshold calculation model calculates the difference and the adaptive threshold according to the detection data and the theoretical data; wherein the threshold calculation module comprises an extreme value data set, a window data set, an initial threshold and an initial deviation value; the initial threshold and the initial deviation data, the initial threshold is obtained according to the extreme value data set, and the initial deviation data is obtained according to the window data set; the extreme value data set is used to calculate the adaptive threshold; The calculated difference is compared with the adaptive threshold, and the health status of the wind turbine tower is evaluated according to the difference between the calculated difference and the adaptive threshold.
2. The wind turbine tower health status assessment method according to claim 1, characterized in that: The wind turbine tower health status assessment method further comprises the step of constructing a relationship model; The steps of constructing the relationship model include: Constructing a modeling data set; wherein the modeling data set includes historical data of the tower in a healthy state, and the historical data includes historical tower top displacement data, historical bending moment data at the connection between adjacent tower sections, and historical unit SCADA data; A relational model is obtained using the modeling data set and based on a DAE algorithm; wherein the relational model is used to output theoretical data corresponding to the input data.
3. The wind turbine tower health status assessment method according to claim 2, characterized in that: In the process of constructing the relational model, the data in the modeling data set is normalized, and the relational model is obtained based on the DAE algorithm using the normalized values.
4. The wind turbine tower health status assessment method according to claim 2, characterized in that: The wind turbine tower health status assessment method further comprises the step of constructing a threshold value calculation model; The step of constructing the threshold calculation model comprises: Calculate historical deviation data based on the modeling data set and modeling theoretical data output by the relational model and corresponding to the modeling data set; An extreme value data set and a window data set are constructed according to the historical deviation data, and the initial threshold is calculated according to the extreme value data set, and the initial deviation data is calculated according to the window data.
5. The wind turbine tower health status assessment method according to claim 4, characterized in that: The step of calculating historical deviation data according to the modeling data set and the modeling theoretical data corresponding to the modeling data set output by the relational model comprises: Obtaining a historical difference value according to the historical bending moment data in the modeling data set and the modeling theoretical bending moment data output by the relationship model and corresponding to the historical bending moment data; converting the historical difference into historical deviation data; The steps of constructing an extreme value data set according to the historical deviation data include: Arrange the plurality of historical deviation data from large to small, and use the value of the t%th data after arrangement as the extreme value threshold value; The data in the historical deviation data that is greater than or equal to the extreme value threshold value is constructed into extreme value data.
6. The wind turbine tower health status assessment method according to claim 5, characterized in that: The formula for converting the historical difference into historical deviation data is: x i =ΔF i -M i ,i=d+1,...,N Among them, x i is the historical deviation data at the i-th time point; M i is the historical deviation average of the d time series before the i-th time point; d is a preset positive integer; N is the number of time points in the modeling data set; ΔF i is the historical difference at the i-th time point.
7. The wind turbine tower health status assessment method according to claim 5, characterized in that: The step of calculating the initial threshold value according to the extreme value data set comprises: x i ′=x i -T PDF: Z q,i =M i +T+b Among them, x′ i is the extreme value data at the i-th time point, x i is the historical deviation data at the i-th time point; T is the extreme value threshold; q is the preset extreme value probability; Z q,i is the threshold value.
8. The wind turbine tower health status assessment method according to claim 4, characterized in that: There are a plurality of historical bending moment data, and the modeling theoretical data includes a plurality of modeling theoretical bending moment data, and the plurality of modeling theoretical bending moment data corresponds one to one with the plurality of historical bending moment data; The step of calculating historical deviation data according to the modeling data set and the modeling theoretical data corresponding to the modeling data set output by the relational model comprises: The difference between the modeling theoretical bending moment data and the corresponding historical bending moment data is calculated, and the largest one of the multiple differences is used as the historical deviation data.
9. A wind turbine, characterized in that: The wind turbine generator includes a tower, a strain sensor, a displacement detection component, a monitoring and control system, and a health status assessment system. The tower includes at least two tower sections. The strain sensor is arranged at the connection between two adjacent tower sections to obtain the bending moment data at the connection between the adjacent tower sections; the displacement detection component is used to obtain the tower top displacement data; the strain sensor, the displacement detection component, and the monitoring and control system are electrically connected to the health status assessment system, the monitoring and control system is used to transmit the unit SCADA data to the health status assessment system, and the health status assessment system is used to execute the wind turbine generator tower health status assessment method described in any one of claims 1 to 8 to perform a health assessment on the tower.
10. The wind turbine generator according to claim 9, characterized in that: The lowest one of the at least two tower sections is a bottom tower section, and the highest one of the at least two tower sections is a top tower section; the displacement detection member comprises a first inclination sensor and a second inclination sensor, the first inclination sensor being arranged on the bottom tower section to obtain an inclination angle of the bottom of the tower; the second inclination sensor being arranged on the top tower section to obtain an inclination angle of the top of the tower; The tower top displacement data is calculated according to the following formula: Among them, d l is the displacement data value of the tower top; l is the setting direction of the first inclination sensor and the second inclination sensor; H is the tower height; θ l,塔顶 is the inclination angle of the tower top; θ l,塔底 It is the inclination angle of the tower bottom in direction l.
Citation Information
Cited By
Method and device for determining health state of wind generating set
CN121542801A
