Fan clearance distance monitoring method and system based on multi-beam radar correlation detection

Abstract

The invention provides a fan clearance distance monitoring method and system based on multi-beam radar correlation detection. The system comprises a three-beam laser radar acquisition module, a data preprocessing module, a consistency verification module, an environment identification and data compensation module, and a clearance value calculation and output module. The method comprises the following steps: S1, extracting echo intensity and slant range data of each beam; s2, performing time sequence filtering processing on each path of acquired beam data, and removing instantaneous jump points deviating from a mean value center; s3, executing multi-beam time correlation verification; s4, executing multi-beam numerical value association verification; s5, calculating a visibility influence factor to judge a visibility environment; and S6, fusing the verified effective data, performing complementation by using the residual effective beam data, and outputting a final fan clearance distance monitoring value. The method is applied to the technical field of wind power generation safety monitoring.

Description

A Method and System for Monitoring Wind Turbine Clearance Distance Based on Multibeam Radar Correlation Detection Technical Field

[0001] This invention belongs to the field of wind power generation safety monitoring technology, specifically relating to a method and system for monitoring the clearance distance of wind turbines based on multi-beam radar correlation detection, which is particularly suitable for accurate monitoring of the clearance distance between the blade tip and the tower wall of wind turbine generators under adverse weather conditions. Background Technology

[0002] The clearance distance of a wind turbine generator, which is the closest distance between the blade tip and the tower wall, is a core critical indicator for ensuring the safe and stable operation of the turbine. Abnormal clearance distance can directly lead to "tower-sweeping" accidents, causing serious damage to the turbine blades and tower, and even resulting in significant losses such as turbine shutdown and power generation interruption. Therefore, real-time and accurate monitoring of the wind turbine clearance distance is a crucial aspect of wind power operation and maintenance.

[0003] Currently, the industry mainly uses single-beam or ordinary multi-beam lidar to detect the clearance distance of wind turbines. However, this type of detection method is highly susceptible to interference from natural environmental factors during actual operation. Specifically, under direct sunlight, fog, rain, or dust storms, the lidar detection signal can be scattered by water droplets, aerosols, dust particles, etc., which not only easily generates a large number of false detection data but also leads to the loss of effective distance measurement data. In particular, during rain or dense fog, water droplets and aerosols can falsely trigger the radar detection signal, causing the radar to identify rain and fog as wind turbine blades, generating a false alarm of "tower sweeping." The wind turbine control system will trigger shutdown protection due to the false alarm, which will seriously reduce the normal power generation efficiency of the wind turbine. In dusty or strong light environments, the effective echo signal of the radar is blocked or interfered with, resulting in the loss of clearance distance data and the inability to achieve effective monitoring, which greatly increases the actual risk of tower sweeping.

[0004] In existing technologies, the interference problem in lidar ranging is often addressed by using single time-domain or spatial-domain filtering for data cleaning. However, these methods can only remove some simple, instantaneous interference and cannot cope with various types of interference under complex and severe weather conditions. Furthermore, there is a lack of effective data completion methods to address the problem of missing effective data after filtering. At the same time, single-beam radars do not have the ability to perform correlation verification, and ordinary multi-beam radars do not utilize the physical logic between beams for signal verification. This makes it difficult to effectively distinguish between the real signals from blades and false signals from rain, fog, dust, etc., resulting in a persistently high false alarm rate.

[0005] In summary, existing wind turbine clearance distance monitoring technologies suffer from poor environmental adaptability, high false alarm rates, and low data reliability, failing to meet the demand for accurate clearance distance monitoring under adverse weather conditions. Therefore, there is an urgent need for a clearance distance monitoring method and system that can effectively eliminate various types of environmental interference, accurately identify false signals, and effectively supplement missing data. Summary of the Invention

[0006] To address the technical problems of poor environmental adaptability, high false alarm rate, and low data reliability in existing wind turbine clearance distance monitoring technologies, this invention provides a wind turbine clearance distance monitoring method and system based on multi-beam radar correlation detection. By combining multi-beam spatial correlation with high repetition rate data filtering, it effectively eliminates instantaneous interference and accurately completes missing data. At the same time, by utilizing multi-beam spatial consistency verification and accurate identification in low-visibility environments, it effectively distinguishes between real blade signals and false environmental signals, solving the problems of false alarms and data loss in severe weather conditions such as rain, dense fog, dust storms, and strong sunlight in existing technologies, and improving the accuracy, reliability, and environmental adaptability of wind turbine clearance distance monitoring.

[0007] The technical solution adopted in this invention is a method for monitoring airspace distance based on multi-beam radar correlation detection. This method includes the following steps: S1, using a multi-beam lidar installed on the wind turbine tower to collect echo signals from at least three laser beams illuminating the blade sweep area at high repetition rate, and extracting the echo intensity and slant range data of each beam; S2, performing time-series filtering on the collected beam data, and using a sliding window standard deviation algorithm to remove instantaneous jump points deviating from the mean center; S3, performing multi-beam time correlation verification, and determining the time when each beam detects the blade signal based on the physical characteristics of blade rotation. S4. Perform multi-beam numerical correlation verification, convert the slant range measured by each beam into clearance distance, and verify whether the converted clearance distance value is within the physically permissible error range; S5. Extract ground detection distance data, calculate visibility impact factor to determine visibility environment, if ground data is lost or abnormal, enter false alarm suppression procedure; S6. Merge the valid data verified by steps S2 to S5, for key beam data missing due to interference, use the remaining valid beam data to complete it through differential prediction model, and output the final wind turbine clearance distance monitoring value.

[0008] Further, in step S1, a three-beam lidar is used to collect the ranging and slant range data of the wind turbine blade tip, and the ranging and slant range data is converted into a horizontal clearance value. The formula for converting the horizontal clearance value is as follows: In the formula, CL i Let L be the horizontal clearance value of the i-th beam. i Let θ be the slant range for the i-th beam.i R is the angle between the i-th beam and the central axis of the tower. tower Let i be the physical radius of the tower at a specific height swept by the blades, i = 1, 2, 3.

[0009] Furthermore, in step S2, a time-space density clustering model is used to clean the data by leveraging the high repetition rate characteristics of the three-beam lidar. Specifically, the echo set of the i-th beam at time t is defined as P. i (t)={d1,d2,…,d n The standard deviation σ of the data within the window is calculated using the sliding window standard deviation formula. window If σ window >σ threshold If the data contains strong interference, then a constant lateral distance distribution filter is initiated; the sliding window standard deviation formula is: The formula for the constant lateral distance filtering is: In the formula, N is the amount of data within the sliding window, and d j For the j-th distance measurement data within the window, σ is the mean of the distance measurement data within the window. threshold d is the standard deviation threshold. ref The weighted predicted value is the blade or ground position identified at the previous moment, where δ is the tolerance and i = 1, 2, 3.

[0010] Furthermore, the three beams of the three-beam lidar are, from the inside out, the first beam, the second beam, and the third beam, and the three beams are directed at the area swept by the blade at different deflection angles.

[0011] Furthermore, in step S3, the multi-beam time correlation verification specifically involves defining the timestamp of the blade passing through each beam as T. i The normally rotating blades must satisfy the time sequence logic of T3→T2→T1 or reverse traversal. If only T1 generates a signal and there is no associated response from T2 and T3, the signal is determined to be an interference signal and is removed.

[0012] Furthermore, in step S4, the multi-beam numerical correlation verification specifically involves defining the mapping clearance function of each beam relative to the tower axis as CL(d i When the blades pass through, the net clearance calculated by each beam must numerically satisfy the following: In the formula, This is the net airspace tolerance threshold.

[0013] Furthermore, in step S5, the calculation of the visibility impact factor η specifically involves: real-time extraction of the ground detection distance D. g The formula for calculating the visibility impact factor η is: In the formula, E echoE represents the ground echo intensity. ref For reference echo intensity, C pts C represents the effective detection rate of ground points. total The total number of ground point detections; the visibility impact factor is set to a critical value of η. critical When η < η critical When the environment is determined to be low visibility, the low visibility recognition mode is activated, and the output of unverified false distance measurement points is prohibited.

[0014] Furthermore, in step S6, when the signal of the first beam is interrupted and missing, and the second and third beams capture stable blade signals, a second-order differential prediction model is used to complete the clearance data of the first beam. The completion formula is as follows: In the formula, CL2(t) is the clearance value of the first beam at time t, CL3(t-Δt) is the clearance value of the second beam at time t, CL3(t-Δt) is the clearance value of the third beam at time t-Δt, and Δt is the time step.

[0015] A wind turbine clearance distance monitoring system for implementing the monitoring method described above includes a three-beam lidar acquisition module, a data preprocessing module, a consistency verification module, an environment identification and data compensation module, and a clearance value calculation and output module. The three-beam lidar acquisition module emits three laser beams with different deflection angles and acquires laser point cloud data from the wind turbine blade tips, extracting echo intensity and ranging slant range. The data preprocessing module converts the ranging slant range into a horizontal clearance value and filters out instantaneous interference noise using a time-space density clustering model. The consistency verification module performs source identification and time-series logic discrimination on the preprocessed clearance value to eliminate false interference signals. The environment identification and data compensation module calculates visibility impact factors to determine low-visibility environments and completes the clearance data for missing beams using a second-order difference prediction model. The clearance value calculation and output module fuses all valid clearance data and outputs the final wind turbine clearance distance value.

[0016] Furthermore, the laser repetition rate of the three-beam lidar acquisition module is 20kHz, and the three beams are distributed radially from the inside to the outside of the tower, corresponding to the inner beam close to the tower, the middle beam, and the outer beam far from the tower, respectively, and the incident angle of each beam is a known geometric installation parameter.

[0017] The beneficial effects of this invention are: 1. High environmental adaptability and effective elimination of various types of environmental interference: This invention combines the high repetition rate characteristics of a three-beam lidar at 20kHz and designs a time-space density clustering signal preprocessing model. First, it identifies instantaneous data jumps caused by rainfall, dust storms, etc., through sliding window standard deviation filtering. Then, it eliminates jump noise points through constant lateral distance distribution filtering, achieving effective filtering of instantaneous interference in the time domain. Simultaneously, it utilizes the constant characteristics of ground echoes to design a visibility influence factor, accurately determining low-visibility environments such as dense fog and strong light, and initiating a false alarm suppression program to prohibit false ranging point outputs, thus avoiding signal misjudgment under low visibility conditions from the environmental identification perspective. This invention, through the combination of time-domain filtering and environmental identification, effectively eliminates interference from various environmental factors such as sunlight, clouds, fog, rainfall, and dust storms, enabling airspace monitoring to operate normally under various severe weather conditions, significantly improving the environmental adaptability of the monitoring technology.

[0018] 2. Multi-dimensional spatial consistency verification significantly reduces false alarm rate: This invention fully utilizes the overlap of the three beams in physical space and the time sequence characteristics of blade sweeping to design a dual spatial consistency verification mechanism of source determination and logical sequence discrimination. On the one hand, the source determination equation constrains the clearance values ​​of different beams to remain highly similar, using the physical characteristics of the blades to eliminate false signals from random scatterers such as rain, fog, and dust. On the other hand, the time sequence logic of blade sweeping constrains the blade signals to appear sequentially according to the order of the beams from the outside to the inside or from the inside to the outside, using the motion trajectory characteristics of the blades to eliminate random interference signals from a single beam. The dual verification mechanism achieves accurate identification of real blade signals and false environmental signals, completely solving the problem of false alarms caused by rain and fog in the prior art, avoiding unplanned shutdowns of wind turbines due to false alarms, and effectively improving the power generation efficiency of wind turbines.

[0019] 3. Multi-beam joint inference for missing data compensation enhances the reliability of monitoring data: Addressing the issue of partial beam signal loss under adverse weather conditions, this invention designs a multi-beam completion algorithm based on a second-order differential prediction model. Utilizing the stable signals of the two outer beams and combining the installation angle differences of each beam, the clearance data of the inner beam closest to the tower is dynamically calculated, achieving accurate data completion. The inner beam is the key beam for monitoring the closest distance between the blade tip and the tower. This invention, through multi-beam joint inference, can still obtain accurate clearance data even when this beam signal is missing. Even if one or two beams are severely interfered with, the system can still maintain effective monitoring through the correlation data of the remaining beams. This ensures the continuity and reliability of clearance monitoring from a data perspective, significantly reducing the risk of tower sweeping due to data loss.

[0020] 4. Modular system design for automated and engineering application of monitoring methods: The monitoring system of this invention is designed with five functional modules corresponding to the monitoring methods. Each module has a clear division of labor and smooth data interaction. It automates steps such as slant range conversion, filtering and cleaning, signal verification, environmental identification, data completion, and clearance output without manual intervention, making it suitable for the on-site operation and maintenance needs of wind turbine generators. At the same time, the installation parameters of the three-beam lidar used in the system are known geometric parameters. Parameters such as beam incident angle and tower radius can be flexibly adjusted according to different wind turbine models, giving it good versatility and engineering promotion value. Attached Figure Description

[0021] Figure 1 is a simplified structural connection diagram of the system of the present invention; Figure 2 is a simplified flowchart of the method of the present invention; Figure 3 is a simplified schematic diagram of the principle of the installation position of the three-beam lidar on the wind turbine. Detailed Implementation

[0022] As shown in Figures 1 to 3, the technical solution of the present invention will be described in detail below with reference to specific embodiments, in order to make the technical content, implementation principle and beneficial effects of the present invention clearer. It should be noted that the embodiments are only used to explain the present invention and are not intended to limit the scope of protection of the present invention.

[0023] A wind turbine clearance distance monitoring system includes a three-beam lidar acquisition module 1, a data preprocessing module 2, a consistency verification module 3, an environment identification and data compensation module 4, and a clearance value calculation and output module 5. The three-beam lidar acquisition module 1 emits three laser beams with different deflection angles and acquires laser point cloud data from the blade tips 8 of the wind turbine 6, extracting echo intensity and ranging slant range. The data preprocessing module 2 converts the ranging slant range into a horizontal clearance value and filters out instantaneous interference noise using a time-space density clustering model. The consistency verification module 3 performs source identification and time-series logic discrimination on the preprocessed clearance value to eliminate false interference signals. The environment identification and data compensation module 4 calculates visibility impact factors to determine low-visibility environments and completes the clearance data for missing beams using a second-order difference prediction model. The clearance value calculation and output module 5 fuses all valid clearance data and outputs the final wind turbine clearance distance value.

[0024] The laser repetition frequency of the three-beam lidar acquisition module 1 is 20kHz. The three beams are distributed radially from the inside to the outside along the tower 7, corresponding to the inner beam close to the tower, the middle beam, and the outer beam far from the tower, respectively. The incident angle of each beam is a known geometric installation parameter.

[0025] The method for monitoring wind turbine clearance distance using the above system includes the following steps: S1. Using a multi-beam lidar installed on the wind turbine tower, at least three laser beams are used to collect echo signals illuminating the blade sweep area at a high repetition rate, and the echo intensity and slant distance data of each beam are extracted; S2. The collected beam data is processed by time series filtering, and instantaneous jump points deviating from the mean center are removed using a sliding window standard deviation algorithm; S3. Multi-beam time correlation verification is performed, and based on the physical characteristics of blade rotation, it is determined whether the timestamps of the signals detected by each beam on the blade meet the preset logical priority. The sequence is as follows: S4, perform multi-beam numerical correlation verification, convert the slant range measured by each beam into clearance distance, and verify whether the converted clearance distance value is within the physically permissible error range; S5, extract the detection distance data of ground 9, calculate the visibility impact factor to determine the visibility environment, and if the ground data is lost or abnormal, enter the false alarm suppression procedure; S6, fuse the valid data verified by steps S2 to S5, and for key beam data missing due to interference, use the remaining valid beam data to complete it through the differential prediction model, and output the final wind turbine clearance distance monitoring value.

[0026] In step S1, a three-beam lidar is used to collect the ranging and slant distance data of the wind turbine blade tip, and the ranging and slant distance data is converted into a horizontal clearance value. The formula for converting the horizontal clearance value is as follows: In the formula, CL i Let L be the horizontal clearance value of the i-th beam. i Let θ be the slant range for the i-th beam. i R is the angle between the i-th beam and the central axis of the tower. tower Let i be the physical radius of the tower at a specific height swept by the blades, i = 1, 2, 3.

[0027] In step S2, a time-space density clustering model is used, and the high repetition rate characteristics of the three-beam lidar are utilized for data cleaning. Specifically, the echo set of the i-th beam at time t is defined as P. i (t)={d1,d2,…,d n The standard deviation σ of the data within the window is calculated using the sliding window standard deviation formula. window If σ window >σ threshold If the data contains strong interference, then a constant lateral distance distribution filter is initiated; the sliding window standard deviation formula is: The formula for the constant lateral distance filtering is: In the formula, N is the amount of data within the sliding window, and d j For the j-th distance measurement data within the window, σ is the mean of the distance measurement data within the window. threshold d is the standard deviation threshold.ref The weighted predicted value is the blade or ground position identified at the previous moment, where δ is the tolerance and i = 1, 2, 3.

[0028] The three beams of the three-beam lidar are, from the inside out, the first beam, the second beam, and the third beam, and the three beams are directed at the area swept by the blade at different deflection angles.

[0029] In step S3, the multi-beam time correlation verification specifically involves defining the timestamp of the blade passing through each beam as T. i The normally rotating blades must satisfy the time sequence logic of T3→T2→T1 or reverse traversal. If only T1 generates a signal and there is no associated response from T2 and T3, the signal is determined to be an interference signal and is removed.

[0030] In step S4, the multi-beam numerical correlation verification specifically involves defining the mapping clearance function of each beam relative to the tower axis as CL(d i When the blades pass through, the net clearance calculated by each beam must numerically satisfy the following: In the formula, This is the net airspace tolerance threshold.

[0031] In step S5, calculating the visibility impact factor η specifically involves: real-time extraction of the ground detection distance D. g The formula for calculating the visibility impact factor η is: In the formula, E echo E represents the ground echo intensity. ref For reference echo intensity, C pts C represents the effective detection rate of ground points. total The total number of ground point detections; the visibility impact factor is set to a critical value of η. critical When η < η critical When the environment is determined to be low visibility, the low visibility recognition mode is activated, and the output of unverified false distance measurement points is prohibited.

[0032] In step S6, when the signal of the first beam is interrupted and missing, and the second and third beams capture stable blade signals, a second-order difference prediction model is used to complete the clearance data of the first beam. The completion formula is as follows: In the formula, CL2(t) is the clearance value of the first beam at time t, CL3(t-Δt) is the clearance value of the second beam at time t, CL3(t-Δt) is the clearance value of the third beam at time t-Δt, and Δt is the time step.

[0033] In a specific embodiment of the present invention, a 2MW doubly-fed wind turbine generator set of a wind farm is selected as the monitoring object. The physical radius R of the tower of the wind turbine at the blade sweep height is... tower=2.5m, the clearance distance is monitored using the three-beam lidar correlation detection system of this invention. The repetition rate of the three-beam lidar is 20kHz. The three beams are distributed radially from the inside to the outside of the tower as the first beam (inner side), the second beam (middle), and the third beam (outer side), with angles of θ1=15°, θ2=20°, and θ3=25° with the central axis of the tower, respectively. The sliding window standard deviation threshold σ is set. threshold =0.3m, lateral distance tolerance δ=0.2m, clearance value tolerance threshold =0.15m, Visibility Influence Factor Critical Value η critical =0.6, time step Δt=5ms.

[0034] Example 1: Monitoring of Wind Turbine Clearance Distance under Normal Weather Conditions. Under clear, windless normal weather conditions, the monitoring method of this invention is activated. The specific execution steps are as follows: S1: The three-beam lidar acquisition module 1 emits three laser beams to acquire laser point cloud data at the tip of the wind turbine blades. The slant distances of the first, second, and third beams are extracted as L1=18.2m, L2=13.5m, and L3=10.8m, respectively, and the echo intensity is stable. S2: The data preprocessing module 2 calculates the horizontal clearance value of each beam using conversion formulas: CL1=18.2×sin15°-2.5≈2.21m, CL2=13.5×sin20°-2.5≈2.23m, CL3=10.8×sin25°-2.5≈2.22m; the sliding window standard deviation σ is calculated. window =0.01m<σ threshold =0.3m, no strong interference, no need to start the lateral distance constant distribution filter; S3, the spatial consistency verification module extracts the timestamps of the blade passing through each beam, satisfying the time series logic of T3→T2→T1; S4, perform multi-beam numerical correlation, △CL 12 =|2.21-2.23|=0.02< =0.15m, the difference in clearance value is within the physically permissible range; S5, extract ground detection distance data and calculate the visibility impact factor η=0.95>η critical =0.6, which is a normal visibility environment, and there is no need to activate false alarm suppression; S6, the airspace value calculation and output module integrates the three effective airspace data, takes the average value and outputs the final airspace distance value as 2.22m.

[0035] In this embodiment, under normal weather conditions, the monitoring method of the present invention can quickly and accurately collect and output airspace distance data, without false signals or missing data, and the monitoring results are highly consistent with the actual airspace distance.

[0036] Example 2: Monitoring of Wind Turbine Clearance Distance in Rainy Weather. In moderate rain, raindrops scatter and interfere with the laser signal. The monitoring method of this invention is activated, and the specific steps are as follows: S1: The ranging and slant range data collected by the three-beam lidar acquisition module 1 exhibit instantaneous jumps. The echo set P1(t) of the first beam shows jump data such as 19.5m and 17.8m, with significant fluctuations in echo intensity; S2: The data preprocessing module 2 calculates the sliding window standard deviation σ. window =0.45m>σ threshold =0.3m, indicating strong rainfall interference, initiate constant lateral distance filtering, and use the weighted predicted value d of the previous moment's leaf position. ref Based on 18.1m, jump data with |d-18.1|≥0.2m were removed, resulting in cleaned slope distances L1=18.2m, L2=13.5m, and L3=10.8m. The converted clearance values ​​are consistent with those in Example 1. The verification results of steps S3 to S6 and subsequent steps are consistent with those in Example 1, and the final output clearance distance value is 2.22m.

[0037] In this embodiment, instantaneous interference noise during rainy weather is effectively eliminated. The monitoring method of the present invention can filter rain interference and output accurate airspace data without generating false alarms.

[0038] Example 3: Monitoring the Clearance Distance of Wind Turbines in Dense Fog. In dense fog, aerosols cause partial signal loss in the first beam, while the signals of the second and third beams remain stable. The monitoring method of this invention is activated, and the specific execution steps are as follows: S1: The three-beam lidar acquisition module 1 acquires the stable ranging slant distances L2=13.5m and L3=10.8m for the second and third beams, respectively. The signal of the first beam is lost due to dense fog interference. S2: The data preprocessing module 2 converts the data to obtain the clearance values ​​CL2=2.23m and CL3=2.22m for the second and third beams, respectively, with no obvious interference noise. S3-S4: The signals of the second and third beams meet the time series logic and the requirement of homology determination. S5: Ground detection distance data is extracted. The ground echo intensity is weakened, and the visibility influence factor η=0.45<η is calculated. critical =0.6, indicating a low visibility dense fog environment, and the low visibility recognition mode is activated; S6, the environment recognition and data compensation module 4 uses a second-order difference prediction model to complete the net clearance data of the first beam. The clearance calculation and output module integrates the supplementary data with the effective data of the second and third beams, and outputs the final clearance distance value of 2.22m.

[0039] In this embodiment, the signal of the inner key beam is missing under dense fog. The monitoring method of the present invention can accurately fill in the missing data through multi-beam joint inference, so as to achieve effective monitoring of the airspace distance.

[0040] Comparative Example 1: Wind Turbine Clearance Monitoring Using Single-Beam LiDAR The same rainy weather environment as in Example 2 was selected. A single-beam lidar in the prior art was used to monitor the clearance distance of the same wind turbine. The parameters of the single-beam lidar were the same as those of the three-beam lidar of the present invention, with only the inner first beam retained.

[0041] Monitoring results showed that the ranging data of the single-beam radar was affected by rainfall, resulting in a large number of instantaneous jumps. The standard deviation of the sliding window reached 0.52m. Although it was filtered by a single time domain, it was still unable to completely eliminate the false signals scattered by raindrops. The radar repeatedly misidentified raindrops as blades, generating 5 false alarms for tower sweeping, which led to 3 unplanned shutdowns of the wind turbines, with a total shutdown time of 1.5 hours, and a significant decrease in power generation efficiency.

[0042] Comparative Example 2: Wind turbine clearance distance monitoring using a conventional multi-beam lidar without data compensation. The same dense fog weather environment as in Example 3 was selected, and a conventional three-beam lidar in the prior art was used to monitor the clearance distance of the same wind turbine. This lidar has a simple spatial filtering function, but it does not have the missing data compensation algorithm of the present invention.

[0043] Monitoring results show that after the dense fog caused the signal of the first beam on the inner side to be lost, ordinary multi-beam radar could only collect the clearance data of the second and third beams, and could not obtain the key data of the closest distance between the blade tip and the tower. It directly output a data loss prompt, and the wind turbine clearance distance monitoring was interrupted, which posed a serious safety risk to the tower sweeping.

[0044] Comparing the above comparative examples and embodiments, it can be seen that: 1. Compared with existing single-beam lidar monitoring technology, the present invention, through multi-beam spatial consistency verification and high repetition rate time domain filtering, can effectively eliminate false signals caused by environmental interference such as rainfall and fog, completely solve the problem of false alarms during tower sweeping, avoid unplanned wind turbine shutdowns, and improve wind turbine power generation efficiency; 2. Compared with existing ordinary multi-beam lidar monitoring technology without data compensation, the missing data compensation algorithm of the present invention can accurately complete the data through multi-beam joint inference when key beam signals are missing, ensuring the continuity of airspace monitoring and eliminating the risk of tower sweeping caused by missing data; 3. The monitoring method and system of the present invention can achieve accurate and continuous monitoring of wind turbine airspace distance under normal and severe weather conditions, and its environmental adaptability and data reliability are far superior to existing technologies, possessing good engineering application value.

[0045] Finally, it should be emphasized that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention can have various changes and modifications. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for monitoring wind turbine clearance distance based on multi-beam radar correlation detection, characterized in that, The method includes the following steps: S1. Using a multi-beam lidar installed on the wind turbine tower, at least three laser beams are used to acquire echo signals illuminating the blade sweep area at a high repetition rate, and the echo intensity and slant range data of each beam are extracted; S2. The acquired beam data is subjected to time series filtering, and instantaneous jump points deviating from the mean center are removed using a sliding window standard deviation algorithm; S3. Multi-beam time correlation verification is performed, and based on the physical characteristics of blade rotation, it is determined whether the timestamps of the signals detected by each beam on the blade meet the preset logical order; S4. Perform multi-beam numerical correlation verification, convert the slant range measured by each beam into clearance distance, and verify whether the converted clearance distance value is within the physically permissible error range; S5, extract ground detection distance data, calculate visibility impact factor to determine visibility environment, if ground data is lost or abnormal, enter false alarm suppression procedure; S6, fuse the valid data verified by steps S2 to S5, for key beam data missing due to interference, use the remaining valid beam data to complete it through differential prediction model, and output the final wind turbine clearance distance monitoring value.

2. The monitoring method according to claim 1, characterized in that, In step S1, a three-beam lidar is used to collect the ranging and slant distance data of the wind turbine blade tip, and the ranging and slant distance data is converted into a horizontal clearance value. The formula for converting the horizontal clearance value is as follows: In the formula, CL i Let L be the horizontal clearance value of the i-th beam. i Let θ be the slant range for the i-th beam. i R is the angle between the i-th beam and the central axis of the tower. tower Let i be the physical radius of the tower at a specific height swept by the blades, i = 1, 2, 3.

3. The monitoring method according to claim 1, characterized in that, In step S2, a time-space density clustering model is used, and the high repetition rate characteristics of the three-beam lidar are utilized for data cleaning. Specifically, the echo set of the i-th beam at time t is defined as P. i (t)={d1,d2,…,d n The standard deviation σ of the data within the window is calculated using the sliding window standard deviation formula. window If σ window >σ threshold If the data contains strong interference, then a constant lateral distance distribution filter is initiated; the sliding window standard deviation formula is: The formula for the constant lateral distance filtering is: In the formula, N is the amount of data within the sliding window, and d j For the j-th distance measurement data within the window, σ is the mean of the distance measurement data within the window. threshold d is the standard deviation threshold. ref The weighted predicted value is the blade or ground position identified at the previous moment, where δ is the tolerance and i = 1, 2, 3.

4. The monitoring method according to claim 2, characterized in that, The three beams of the three-beam lidar are, from the inside out, the first beam, the second beam, and the third beam, and the three beams are directed at the area swept by the blade at different deflection angles.

5. The monitoring method according to claim 1, characterized in that, In step S3, the multi-beam time correlation verification specifically involves defining the timestamp of the blade passing through each beam as T. i The normally rotating blades must satisfy the time sequence logic of T3→T2→T1 or reverse traversal. If only T1 generates a signal and there is no associated response from T2 and T3, the signal is determined to be an interference signal and is removed.

6. The monitoring method according to claim 1, characterized in that, In step S4, the multi-beam numerical correlation verification specifically involves defining the mapping clearance function of each beam relative to the tower axis as CL(d i When the blades pass through, the net clearance calculated by each beam must numerically satisfy the following: In the formula, This is the net airspace tolerance threshold.

7. The monitoring method according to claim 1, characterized in that, In step S5, calculating the visibility impact factor η specifically involves: real-time extraction of the ground detection distance D. g The formula for calculating the visibility impact factor η is: In the formula, E echo E represents the ground echo intensity. ref For reference echo intensity, C pts C represents the effective detection rate of ground points. total The total number of ground point detections; the visibility impact factor is set to a critical value of η. critical When η < η critical When the environment is determined to be low visibility, the low visibility recognition mode is activated, and the output of unverified false distance measurement points is prohibited.

8. The monitoring method according to claim 4, characterized in that, In step S6, when the signal of the first beam is interrupted and missing, and the second and third beams capture stable blade signals, a second-order difference prediction model is used to complete the clearance data of the first beam. The completion formula is as follows: In the formula, CL2(t) is the clearance value of the first beam at time t, CL3(t-Δt) is the clearance value of the second beam at time t, CL3(t-Δt) is the clearance value of the third beam at time t-Δt, and Δt is the time step.

9. A wind turbine clearance distance monitoring system for implementing the monitoring method as described in any one of claims 1 to 8, characterized in that, The monitoring system includes a three-beam lidar acquisition module (1), a data preprocessing module (2), a consistency verification module (3), an environment identification and data compensation module (4), and a headroom calculation and output module (5). The three-beam lidar acquisition module (1) is used to emit three laser beams with different deflection angles and acquire laser point cloud data of the wind turbine blade tip, and extract echo intensity and ranging slant distance. The data preprocessing module (2) is used to convert the ranging slant distance into a horizontal headroom value and filter out instantaneous interference noise through a time-space density clustering model. The consistency verification module (3) is used to perform homogeneity determination and time series logic discrimination on the preprocessed headroom value to eliminate false interference signals. The environment identification and data compensation module (4) is used to calculate the visibility impact factor to determine the low visibility environment and complete the headroom data of the missing beams through a second-order difference prediction model. The headroom calculation and output module (5) is used to fuse all valid headroom data and output the final wind turbine headroom distance value.

10. The monitoring system according to claim 9, characterized in that, The laser repetition frequency of the three-beam lidar acquisition module (1) is 20kHz. The three beams are distributed radially from the inside to the outside of the tower, corresponding to the inner beam close to the tower, the middle beam, and the outer beam far from the tower, respectively. The incident angle of each beam is a known geometric installation parameter.

Images