Clearance radar lens smudginess detection method, device and equipment and storage medium

Abstract

The invention discloses a clearance radar lens smudginess detection method, device and equipment and a storage medium, and is applied to the field of wind power generation. Various sensor data are acquired in a current detection period; distance measurement data acquired by the wind turbine generator clearance radar in the current detection period are acquired, and an abnormal value proportion is determined based on the proportion of lens smudginess abnormal return values in the distance measurement data; determining a compensation coefficient of each sensor based on the sensor data, and determining a working condition weight of each sensor based on the smudginess state contribution degree of the sensor data; determining a correction coefficient based on the working condition weight and the compensation coefficient; and performing correction based on the correction coefficient to obtain a target abnormal value ratio, and determining a clearance radar lens smudginess detection result based on the target abnormal value ratio. According to the invention, automatic detection of the smudginess of the clearance radar lens is carried out by collecting sensor data, and the problems of low efficiency, high false alarm rate and serious waste of operation and maintenance resources caused by manual regular detection, cleaning and maintenance in the prior art are avoided.

Description

Technical Field

[0001] This invention relates to the field of wind power generation, and in particular to a method for detecting dirt in a radar lens, a device for detecting dirt in a radar lens, an electronic device, and a computer-readable storage medium. Background Technology

[0002] Blade-tower sweeping incidents in wind turbines can lead to structural damage to the blades and tower, and even tower collapse, causing significant economic losses. Clearance radar technology, by monitoring the blade-tower distance in real time, can effectively prevent sweeping incidents in existing turbines and eliminate power limitations caused by safety concerns, significantly increasing power generation. In newer turbines, it can optimize blade design redundancy and reduce the load design pressure on the turbine. However, clearance radar lenses are exposed to complex environments for extended periods, and contaminant accumulation degrades their performance. Current maintenance methods, such as periodic cleaning, lack specificity and risk over- or under-maintenance, resulting in low efficiency, high false alarm rates, and significant waste of maintenance resources. Summary of the Invention

[0003] The purpose of this invention is to provide a method, device, equipment, and storage medium for detecting dirt on a radar lens, which can be applied to the field of wind power generation. This method automatically detects dirt on the radar lens by collecting sensor data, avoiding the problems of low efficiency, high false alarm rate, and serious waste of operation and maintenance resources caused by manual periodic detection, cleaning and maintenance in the prior art.

[0004] To solve the above-mentioned technical problems, the present invention provides a method for detecting dirt in an air clearance radar lens, comprising:

[0005] During the current detection cycle, multiple sensor data are collected based on multiple sensors; the multiple sensor data include wind turbine environmental data and wind turbine operation data;

[0006] Acquire ranging data collected by the wind turbine's airspace radar during the current detection cycle, and determine the proportion of abnormal values ​​based on the proportion of abnormal return values ​​for lens dirt in the ranging data.

[0007] The compensation coefficient of each sensor is determined based on the sensor data, and the operating condition weight of each sensor is determined based on the contribution of the dirt status of the sensor data.

[0008] The correction coefficient for the proportion of outliers is determined based on the operating condition weights and compensation coefficients of each of the aforementioned sensors.

[0009] The target outlier percentage is obtained by correcting the outlier percentage based on the correction coefficient, and the dirt detection result of the airspace radar lens is determined based on the target outlier percentage.

[0010] Optionally, the correction coefficient for determining the proportion of outliers based on the operating condition weights of each sensor and the compensation coefficient includes:

[0011] The initial correction coefficient for the proportion of outliers in the current detection cycle is determined based on the operating condition weights and compensation coefficients of each sensor.

[0012] Obtain the historical correction coefficient from the previous detection period, determine the forgetting factor, and then perform a weighted summation of the initial correction coefficient and the historical correction coefficient based on the forgetting factor to obtain the correction coefficient for the current detection period.

[0013] Optionally, the method further includes:

[0014] The data from multiple sensors collected in each historical detection period are defined as a sample, and the correction coefficient for the proportion of outliers in the historical detection period is defined as the sample label.

[0015] Based on the samples and their labels, a dataset prediction model with corrected coefficients is trained.

[0016] The various sensor data obtained during the current detection period are identified as target data, and the target data are input into the trained correction coefficient prediction model to obtain the output correction coefficient.

[0017] The target outlier percentage is obtained by correcting the outlier percentage in the current detection period based on the correction coefficient.

[0018] Optionally, the operating condition weight of each sensor is determined based on the contribution of the contamination status of the sensor data, including:

[0019] Determine the information entropy of each sensor, and determine the contribution of the dirt state of the sensor data based on the information entropy;

[0020] The total dirt state contribution is obtained by summing the dirt state contribution values ​​of each sensor data.

[0021] The proportion of the contribution of the dirt state of the sensor data to the total contribution of dirt state is determined as the operating condition weight of the sensor.

[0022] Optionally, the dirt detection result of the airspace radar lens is determined based on the proportion of the target outliers, including:

[0023] Set up a state machine that includes multiple lens dirt states, determine the state transition rules for each lens dirt state based on the outlier percentage threshold, and set the execution actions after each lens dirt state undergoes a state transition.

[0024] The dirt detection result of the air clearance radar lens is obtained by detecting the proportion of outliers of the target based on the state machine.

[0025] Optionally, the proportion of outliers is determined based on the proportion of lens dirt abnormality return values ​​in the ranging data, including:

[0026] The distance measurement data is obtained by deleting all data from the initial distance measurement data except for the lens dirt abnormal return value and the normal distance measurement value; the lens dirt abnormal return value is 65535.

[0027] The percentage of abnormal return values ​​for lens dirt in the total data volume of the ranging data is determined as the percentage of abnormal values.

[0028] Optionally, the various sensors include: wind speed and direction sensors, temperature and humidity sensors, vibration monitoring sensors, rainfall measurement sensors, and ambient light sensors.

[0029] To solve the above-mentioned technical problems, the present invention provides a dirt detection device for an air defense radar lens, comprising:

[0030] The first module is used to collect various sensor data based on multiple sensors within the current detection cycle; the various sensor data include wind turbine environmental data and wind turbine operation data.

[0031] The second module is used to acquire ranging data collected by the wind turbine's airspace radar during the current detection cycle, and to determine the proportion of abnormal values ​​based on the proportion of abnormal return values ​​for lens dirt in the ranging data.

[0032] The third module is used to determine the compensation coefficient of each sensor based on the sensor data, and to determine the operating condition weight of each sensor based on the contribution of the dirt status of the sensor data.

[0033] The fourth module is used to determine the correction coefficient for the proportion of outliers based on the operating condition weights of each sensor and the compensation coefficient.

[0034] The fifth module is used to correct the outlier ratio based on the correction coefficient to obtain the target outlier ratio, and to determine the dirt detection result of the airspace radar lens based on the target outlier ratio.

[0035] To solve the above-mentioned technical problems, the present invention provides an electronic device, comprising:

[0036] Memory, used to store computer programs;

[0037] A processor is used to implement the above-described method for detecting dirt in a radar lens when executing the computer program.

[0038] To solve the above-mentioned technical problems, the present invention provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the above-described method for detecting dirt in a radar lens.

[0039] As can be seen, this invention acquires multiple sensor data based on multiple sensors within the current detection cycle; the multiple sensor data includes wind turbine environmental data and wind turbine operation data; it acquires ranging data collected by the wind turbine air clearance radar within the current detection cycle, determines the proportion of outliers based on the proportion of lens contamination abnormal return values ​​in the ranging data; it determines the compensation coefficient of each sensor based on the sensor data, and determines the operating condition weight of each sensor based on the contribution of the contamination status of the sensor data; it determines the correction coefficient of the outlier proportion based on the operating condition weight and compensation coefficient of each sensor; it corrects the outlier proportion based on the correction coefficient to obtain the target outlier proportion, and determines the air clearance radar lens contamination detection result based on the target outlier proportion.

[0040] This invention automates the detection of dirt on the radar lens of the airspace by collecting sensor data, avoiding the problems of low efficiency, high false alarm rate and serious waste of operation and maintenance resources caused by manual periodic detection, cleaning and maintenance in the prior art. Attached Figure Description

[0041] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0042] Figure 1 A flowchart of a method for detecting dirt in an air clearance radar lens provided in an embodiment of the present invention;

[0043] Figure 2 This is an example diagram of a state machine structure provided in an embodiment of the present invention;

[0044] Figure 3 This is a structural block diagram of a dirt detection device for an air clearance radar lens provided in an embodiment of the present invention. Detailed Implementation

[0045] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0046] During the operation of wind turbine generators, dynamic conditions such as turbulence and extreme wind speeds, coupled with the trend of larger blades (such as increased length and intensified flexible deformation), can easily lead to blade-tower sweep risks. Such incidents can cause structural damage to the blades and towers at best, and tower collapse at worst, resulting in significant economic losses for wind farm operators and turbine manufacturers.

[0047] Air clearance radar technology can effectively prevent tower sweeping accidents and eliminate power limitation caused by safety concerns in existing units by monitoring the blade-to-tower distance in real time, thus significantly increasing power generation. In new units, it can optimize blade design redundancy and reduce the load design pressure on the unit.

[0048] Airspace clearance radar lenses are exposed to complex environments such as wind, sand, rain, snow, and salt spray for extended periods. Contaminant deposits can significantly attenuate laser signal transmission and reception, leading to reduced or even complete loss of the radar's ranging and reflectivity measurement capabilities, directly threatening the reliability of monitoring the safe operation of the aircraft. Therefore, to ensure the normal operation of the airspace clearance radar, it is necessary to regularly check for dirt on the radar lenses and promptly alert maintenance personnel to remove any dirt found.

[0049] Current mainstream solutions in the industry have obvious shortcomings: regular cleaning (usually every 3 months) maintenance lacks specificity and carries the risk of over-maintenance or under-maintenance; traditional methods for detecting dirt in airspace radar lenses have low data efficiency and high false alarm rates, resulting in a serious waste of operation and maintenance resources.

[0050] The following combination Figure 1 , Figure 1 A flowchart of a method for detecting dirt in a radar lens provided in an embodiment of the present invention, the method may include:

[0051] S101: Multiple sensor data are collected based on multiple sensors during the current detection cycle; the multiple sensor data includes wind turbine environmental data and wind turbine operation data.

[0052] In this embodiment, to achieve high-precision intelligent identification of dirt on the airspace radar lens, a multi-source heterogeneous sensor data acquisition network can be constructed first. Multi-dimensional environmental and unit status data can be acquired through a high-precision sensor array, and the acquired data can be transmitted in real time to the SCADA (Supervisory Control And Data Acquisition) system data platform.

[0053] This embodiment does not limit the sampling frequency of the sensor, nor does it limit the specific types of various sensors; the settings can be based on the actual application. Generally, the sampling frequency of the sensor can be 1Hz, and various sensors can include: wind speed and direction sensors, temperature and humidity sensors, vibration monitoring sensors, rainfall measurement sensors, and ambient light sensors.

[0054] Wind speed and direction sensors can be used to obtain the real-time wind speed and direction of the current unit; temperature and humidity sensors can be used to obtain the current ambient temperature and humidity; vibration monitoring sensors can be used to monitor the current vibration of the unit; rainfall sensors can be used to monitor the current weather conditions; and ambient light sensors can be used to monitor ambient visibility.

[0055] This embodiment allows setting the detection cycle for dirt detection on the air clearance radar lens. The duration of the detection cycle can be set based on actual applications, typically 24 hours. Within the current detection cycle, this embodiment can collect various sensor data, including wind turbine environmental data and wind turbine operating data.

[0056] S102: Obtain the ranging data collected by the wind turbine's airspace radar during the current detection cycle, and determine the proportion of abnormal values ​​based on the proportion of abnormal return values ​​for lens dirt in the ranging data.

[0057] In this embodiment, the clearance radar can employ three independently calibrated 905nm laser beams to achieve millimeter-level precision measurement of the clearance distance between the wind turbine blade and the tower, as well as prediction of distance change trends, by detecting the clearance distance at three different positions on the blade. The normal ranging range of the clearance radar is 0~15000cm (corresponding to 0-150m, covering mainstream tower heights), and the sampling frequency of the clearance radar ranging data can be synchronized with the sensor sampling frequency.

[0058] Air clearance radar technology can effectively prevent tower sweeping accidents by monitoring the distance between the blades and the tower in real time. In this embodiment, the ranging data collected by the air clearance radar of the wind turbine in the current detection cycle can be obtained, and the proportion of abnormal values ​​can be determined based on the proportion of abnormal return values ​​of lens dirt in the ranging data.

[0059] Lens dirt abnormal return value generally refers to abnormal return value in the ranging data that is related to lens dirt, such as the 65535 return value. Therefore, in this embodiment, the proportion of abnormal values ​​can be determined based on the proportion of lens dirt abnormal return values ​​in the ranging data, that is, the proportion of lens dirt abnormal return values ​​in the ranging data.

[0060] However, since there may be other abnormal return values ​​in the ranging data besides the 65535 return value, this embodiment can delete the data other than the abnormal return value of lens dirt and the normal ranging value in the initial ranging data to obtain the ranging data; the proportion of abnormal return values ​​of lens dirt in the total data volume of the ranging data is determined as the abnormal value proportion.

[0061] S103: Determine the compensation coefficient of each sensor based on sensor data, and determine the operating condition weight of each sensor based on the contribution of the dirt status of sensor data.

[0062] S104: Correction coefficient for determining the proportion of outliers based on the operating condition weights and compensation coefficients of each sensor.

[0063] Because the 65535 return value has multiple causes, typical operating conditions for abnormal 65535 return values ​​include: lens dirt; communication abnormalities: hardware failures such as loose connectors; environmental interference: rainy or foggy weather, sandstorms, freezing, etc.; unit vibration; biological interference: birds flying over the detection area.

[0064] In other words, apart from lens dirt, the 65535 return value can be generated by other operating conditions. Therefore, this embodiment can use cross-validation of multiple sensor data to determine the correction coefficient for the proportion of outliers.

[0065] This embodiment does not limit the specific method for determining the correction coefficient for the proportion of outliers. Generally, the compensation coefficient of each sensor can be determined based on the sensor data, the operating weight of each sensor can be determined based on the contribution of the dirt status of the sensor data, and the correction coefficient for the proportion of outliers can be determined based on the operating weight and compensation coefficient of each sensor.

[0066] This embodiment does not limit the specific method of determining the compensation coefficient of each sensor. Generally, the compensation coefficient of the corresponding sensor can be obtained by analyzing the sensor data. For example, in this embodiment, the wind speed compensation coefficient of the wind speed and wind direction sensor, the vibration compensation coefficient of the vibration monitoring sensor, the freezing compensation coefficient of the temperature and humidity sensor, the rain and fog compensation coefficient of the rainfall metering sensor, and the sandstorm compensation coefficient of the ambient light sensor can be determined.

[0067] In this embodiment, the compensation coefficient of each sensor can be set within a certain range. For example, the wind speed compensation coefficient can be in the range of 0.8 to 1 (inversely proportional to the square of the wind speed); the vibration compensation coefficient can be in the range of 0.8 to 1 (linearly related to the vibration amplitude); the freezing compensation coefficient can be in the range of 0.8 to 1 (calculated based on the temperature gradient); the rain and fog compensation coefficient can be in the range of 0.8 to 1 (related to the relative humidity index); and the sandstorm compensation coefficient can be in the range of 0.8 to 1 (determined based on the visibility model).

[0068] This embodiment can determine the operating weight of each sensor based on the contribution of the dirt state to the sensor data. This embodiment does not limit the specific method of determining the contribution of the dirt state. It can be determined by the compensation coefficient of the sensor. For example, the larger the compensation coefficient, the higher the contribution of the dirt state and the higher its corresponding operating weight.

[0069] This embodiment can also determine the contribution of the contamination state through information entropy. Specifically, firstly, the information entropy of each sensor can be determined, and the contribution of the sensor data to the contamination state can be determined based on the information entropy. Then, the contribution of the contamination state of each sensor data is summed to obtain the total contribution of the contamination state. The proportion of the contribution of the sensor data to the total contribution of the contamination state is determined as the operating condition weight of the sensor.

[0070] The calculation method for load condition weights can be shown in the following formula:

[0071] ;

[0072] In the formula, W i E represents the operating condition weight of the i-th sensor. i Let E be the information entropy of the i-th sensor. k Let be the information entropy of the k-th sensor, and n be the total number of sensors, 1-E i The contribution of the dirt status to the i-th sensor.

[0073] Each compensation coefficient can correspond to a working condition weight. In this embodiment, the correction coefficient for the proportion of outliers can be determined based on the working condition weights and compensation coefficients of each sensor.

[0074] This embodiment does not limit the specific method for determining the correction coefficient. Generally, the compensation coefficient can be weighted and summed using operating condition weights, and the summation result can be divided by the total number of sensors, as shown in the following formula:

[0075] ;

[0076] In the formula, T is the correction coefficient, n is the total number of sensors, and W i Let D be the operating condition weight for the i-th sensor. i Let be the compensation coefficient for the i-th sensor.

[0077] In this embodiment, considering the correlation between the sensor data from the previous detection cycle and the sensor data from the current detection cycle, the initial correction coefficient T for the proportion of outliers in the current detection cycle is determined based on the operating condition weights and compensation coefficients of each sensor. origin Obtain the historical correction coefficient T from the previous detection period. historyThe forgetting factor δ is determined, and the initial correction coefficient and historical correction coefficients are weighted and summed based on the forgetting factor to obtain the correction coefficient T for the current detection period, as shown in the following formula:

[0078] T=(1-δ)T origin +δT history .

[0079] S105: The outlier ratio is corrected based on the correction coefficient to obtain the target outlier ratio, and the dirt detection result of the airspace radar lens is determined based on the target outlier ratio.

[0080] This embodiment can obtain the target outlier ratio by correcting the outlier ratio based on the correction coefficient, as shown in the following formula:

[0081] P=T*ω;

[0082] In the formula, P is the target anomaly ratio, ω is the outlier ratio, and T is the correction coefficient.

[0083] After the correction is completed, this embodiment can determine the dirt detection result of the airspace radar lens based on the proportion of outliers in the target. This embodiment does not limit the specific method of determining the dirt detection result of the airspace radar lens. Generally, various lens dirt states can be set, and each lens dirt state can correspond to a dirt level. The dirt level can be divided by setting an outlier percentage threshold. For example, the outlier percentage threshold for the normal state can be set to 0~10%. When the target outlier percentage is within the outlier percentage threshold range of the normal state, it can be determined that the lens is in normal condition and is free of dirt.

[0084] Specifically, this embodiment can achieve lens contamination detection by setting a state machine. A state machine containing multiple lens contamination states can be set, and the state transition rules for each lens contamination state are determined based on the outlier percentage threshold. The execution actions after the state transition of each lens contamination state are set. The lens contamination detection result of the air clearance radar is obtained by detecting the outlier percentage of the target based on the state machine.

[0085] An example of the overall architecture diagram of a state machine can be found as follows: Figure 2 As shown, the lens dirt status can include: normal status, warning level status, alarm level status and fault level status, and each lens dirt status corresponds to an abnormal value percentage threshold.

[0086] For example, when the target anomaly percentage P < 10%, the airspace radar is in normal condition; when the target anomaly percentage 10% ≤ P < 20%, the event is recorded and a "suspected minor dirtiness warning for airspace radar lens" is output to the SCADA system, i.e., warning level; when the target anomaly percentage recovers to P < 10% after a period of time, the "suspected minor dirtiness warning for airspace radar lens" record is cleared, and the airspace radar is in normal condition; when the target anomaly percentage 20% ≤ P < 30%, the event is recorded and a "moderate dirtiness alarm for airspace radar lens" is output to the SCADA system, i.e., alarm level; when the target anomaly percentage recovers to 10% ≤ P < 20% after a period of time, the "moderate dirtiness alarm for airspace radar lens" record is cleared, and the airspace radar is in warning level; when the target anomaly percentage P ≥ 30%, the event is recorded, the system is stopped and locked, and a "severe dirtiness fault for airspace radar lens" is output to the SCADA system, i.e. fault level; then, manual cleaning is notified, the system is reset, and relevant data for this fault are recorded.

[0087] Since determining the correction coefficient in practical applications requires a large amount of computing resources, which is inefficient and costly, this embodiment can introduce a neural network model to replace the calculation process of the correction coefficient.

[0088] Specifically, in this embodiment, multiple sensor data collected in each historical detection cycle can be used as a sample, and the correction coefficient of the proportion of outliers in the historical detection cycle can be determined as the sample label.

[0089] The modified coefficient prediction model is trained based on the dataset constructed from the samples and sample labels. This embodiment does not limit the specific type of the modified coefficient prediction model, but it can generally be an LSTM (Long Short-Term Memory) model.

[0090] The input layer of the LSTM model has a dimension of 5 (corresponding to 5 sensor features); its architecture can adopt a two-layer LSTM design. The first layer can contain 96 units to capture short-term patterns (<30 seconds), such as sudden oil splashes; the second layer can contain 48 units to learn long-term dependencies (>5 minutes), such as slow dust accumulation; the Dropout layer can be set with a dropout rate of 0.2~0.3 to prevent overfitting; the output layer outputs the predicted correction coefficients.

[0091] Multiple sensor data obtained in the current detection period are used as target data. The target data is input into the trained correction coefficient prediction model to obtain the output correction coefficient. The outlier ratio in the current detection period is corrected based on the correction coefficient to obtain the target outlier ratio.

[0092] Based on the above embodiments, the present invention automatically detects dirt on the radar lens by collecting sensor data, avoiding the problems of low efficiency, high false alarm rate and serious waste of operation and maintenance resources caused by manual periodic detection, cleaning and maintenance in the prior art.

[0093] The following combination Figure 3 , Figure 3 This is a structural block diagram of a dirt detection device for an air clearance radar lens provided in an embodiment of the present invention. The device may include:

[0094] The first module 100 is used to collect various sensor data based on multiple sensors during the current detection cycle; the various sensor data include wind turbine environmental data and wind turbine operation data.

[0095] The second module 200 is used to acquire the ranging data collected by the wind turbine's airspace radar during the current detection cycle, and to determine the proportion of abnormal values ​​based on the proportion of abnormal return values ​​for lens dirt in the ranging data.

[0096] The third module 300 is used to determine the compensation coefficient of each sensor based on sensor data, and to determine the operating condition weight of each sensor based on the contribution of the dirt status of sensor data.

[0097] The fourth module 400 is used to determine the correction coefficient for the proportion of outliers based on the operating condition weights and compensation coefficients of each sensor.

[0098] The fifth module 500 is used to correct the outlier ratio based on the correction coefficient to obtain the target outlier ratio, and to determine the dirt detection result of the airspace radar lens based on the target outlier ratio.

[0099] Based on the above embodiments, the present invention automatically detects dirt on the radar lens by collecting sensor data, avoiding the problems of low efficiency, high false alarm rate and serious waste of operation and maintenance resources caused by manual periodic detection, cleaning and maintenance in the prior art.

[0100] Based on the above embodiments, the fourth module 400 may include:

[0101] The first unit is used to determine the initial correction coefficient for the proportion of outliers in the current detection cycle based on the operating condition weights and compensation coefficients of each sensor.

[0102] The second unit is used to obtain the historical correction coefficients of the previous detection period, determine the forgetting factor, and then perform a weighted summation of the initial correction coefficients and historical correction coefficients based on the forgetting factor to obtain the correction coefficients for the current detection period.

[0103] Based on the above embodiments, the device may further include:

[0104] The sixth module is used to identify multiple sensor data collected in each historical detection period as a sample, and to determine the correction coefficient of the proportion of outliers in the historical detection period as the sample label.

[0105] The seventh module is used to build a dataset based on samples and sample labels to train a prediction model with corrected coefficients;

[0106] The eighth module is used to identify multiple sensor data obtained in the current detection period as target data, and input the target data into the trained correction coefficient prediction model to obtain the output correction coefficient.

[0107] The ninth module is used to correct the proportion of outliers in the current detection period based on the correction coefficient to obtain the target proportion of outliers.

[0108] Based on the above embodiments, the third module 300 may include:

[0109] The third unit is used to determine the information entropy of each sensor and, based on the information entropy, to determine the contribution of the sensor data to the state of contamination.

[0110] The fourth unit is used to sum the dirt state contribution of each sensor data to obtain the total dirt state contribution.

[0111] The fifth unit is used to determine the proportion of the sensor data's contribution to the overall contribution to the overall contamination status as the sensor's operating condition weight.

[0112] Based on the above embodiments, the fifth module 500 may include:

[0113] The sixth unit is used to set up a state machine that includes multiple lens dirt states, determine the state transition rules for each lens dirt state based on the outlier percentage threshold, and set the execution actions after each lens dirt state undergoes a state transition.

[0114] The seventh unit is used to detect the proportion of outliers in the target based on the state machine to obtain the dirt detection results of the airspace radar lens.

[0115] Based on the above embodiments, the second module 200 may include:

[0116] The eighth unit is used to delete all data in the initial ranging data except for the lens dirt abnormal return value and the normal ranging value to obtain the ranging data; the lens dirt abnormal return value is 65535.

[0117] The ninth unit is used to determine the percentage of abnormal return values ​​due to lens dirt in the total data volume of the ranging data as the percentage of abnormal values.

[0118] Based on the above embodiments, various sensors include: wind speed and direction sensors, temperature and humidity sensors, vibration monitoring sensors, rainfall measurement sensors, and ambient light sensors.

[0119] Based on the above embodiments, the present invention also provides an electronic device, which may include a memory and a processor. The memory stores a computer program, and when the processor calls the computer program in the memory, it can implement the steps provided in the above embodiments. Of course, the device may also include various necessary network interfaces, a power supply, and other components.

[0120] The present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by an execution terminal or processor, can implement the method provided in the embodiments of the present invention; the storage medium may include various media capable of storing program code, such as a USB flash drive, a portable hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0121] In this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, without necessarily requiring or implying any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

Claims

1. A method for detecting dirt and grime in a radar lens, characterized in that, include: Multiple sensor data are collected based on multiple sensors during the current detection cycle; The various sensor data include wind turbine environmental data and wind turbine operating data; Acquire ranging data collected by the wind turbine's airspace radar during the current detection cycle, and determine the proportion of abnormal values ​​based on the proportion of abnormal return values ​​for lens dirt in the ranging data. The compensation coefficient of each sensor is determined based on the sensor data, and the operating condition weight of each sensor is determined based on the contribution of the dirt status of the sensor data. The correction coefficient for the proportion of outliers is determined based on the operating condition weights and compensation coefficients of each of the aforementioned sensors. The target outlier percentage is obtained by correcting the outlier percentage based on the correction coefficient, and the dirt detection result of the airspace radar lens is determined based on the target outlier percentage.

2. The method for detecting dirt in a radar lens according to claim 1, characterized in that, The correction coefficient for determining the proportion of outliers based on the operating condition weights and compensation coefficients of each sensor includes: The initial correction coefficient for the proportion of outliers in the current detection cycle is determined based on the operating condition weights and compensation coefficients of each sensor. Obtain the historical correction coefficient from the previous detection period, determine the forgetting factor, and then perform a weighted summation of the initial correction coefficient and the historical correction coefficient based on the forgetting factor to obtain the correction coefficient for the current detection period.

3. The method for detecting dirt in a radar lens according to claim 1, characterized in that, Also includes: The data from multiple sensors collected in each historical detection period are defined as a sample, and the correction coefficient for the proportion of outliers in the historical detection period is defined as the sample label. Based on the samples and their labels, a dataset prediction model with corrected coefficients is trained. The various sensor data obtained during the current detection period are identified as target data, and the target data are input into the trained correction coefficient prediction model to obtain the output correction coefficient. The target outlier percentage is obtained by correcting the outlier percentage in the current detection period based on the correction coefficient.

4. The method for detecting dirt in a radar lens according to claim 1, characterized in that, The operating condition weight of each sensor is determined based on the contribution of the dirt status to the sensor data, including: Determine the information entropy of each sensor, and determine the contribution of the dirt state of the sensor data based on the information entropy; The total dirt state contribution is obtained by summing the dirt state contribution values ​​of each sensor data. The proportion of the contribution of the dirt state of the sensor data to the total contribution of dirt state is determined as the operating condition weight of the sensor.

5. The method for detecting dirt in a radar lens according to claim 1, characterized in that, The airspace radar lens dirt detection result is determined based on the proportion of outliers in the target, including: Set up a state machine that includes multiple lens dirt states, determine the state transition rules for each lens dirt state based on the outlier percentage threshold, and set the execution actions after each lens dirt state undergoes a state transition. The dirt detection result of the air clearance radar lens is obtained by detecting the proportion of outliers of the target based on the state machine.

6. The method for detecting dirt in a radar lens according to claim 1, characterized in that, The proportion of outliers is determined based on the percentage of abnormal lens contamination values ​​returned in the ranging data, including: The distance measurement data is obtained by deleting all data from the initial distance measurement data except for the lens dirt abnormal return value and the normal distance measurement value; the lens dirt abnormal return value is 65535. The percentage of abnormal return values ​​for lens dirt in the total data volume of the ranging data is determined as the percentage of abnormal values.

7. The method for detecting dirt in a radar lens according to claim 1, characterized in that, The various sensors include: wind speed and direction sensors, temperature and humidity sensors, vibration monitoring sensors, rainfall measurement sensors, and ambient light sensors.

8. A device for detecting dirt and grime in a radar lens, characterized in that, include: The first module is used to collect data from multiple sensors within the current detection cycle. The various sensor data include wind turbine environmental data and wind turbine operating data; The second module is used to acquire ranging data collected by the wind turbine's airspace radar during the current detection cycle, and to determine the proportion of abnormal values ​​based on the proportion of abnormal return values ​​for lens dirt in the ranging data. The third module is used to determine the compensation coefficient of each sensor based on the sensor data, and to determine the operating condition weight of each sensor based on the contribution of the dirt status of the sensor data. The fourth module is used to determine the correction coefficient for the proportion of outliers based on the operating condition weights of each sensor and the compensation coefficient. The fifth module is used to correct the outlier ratio based on the correction coefficient to obtain the target outlier ratio, and to determine the dirt detection result of the airspace radar lens based on the target outlier ratio.

9. An electronic device, characterized in that, include: Memory, used to store computer programs; A processor, configured to implement the air clearance radar lens dirt detection method as described in any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, implement the air clearance radar lens dirt detection method as described in any one of claims 1 to 7.

Images