Online Monitoring System, Method, Electronic Device and Medium for Wind Turbine Blades

Abstract

The present invention provides an on-line monitoring system, method, electronic device and medium for wind turbine blades, belonging to the field of wind turbine testing. The system includes: a low-frequency vibration sensor installed at a position on the wind turbine blade at a preset distance from the blade root for collecting vibration signals of the wind turbine blade; a CANopen communication module for receiving and transmitting the vibration signals of the wind turbine blade; a CANopen master station terminal module for transmitting the vibration signals of the wind turbine blade to the main control system; and a main control system for processing and analyzing the blade vibration signals through an embedded fault diagnosis algorithm to realize real-time monitoring of the wind turbine blade. By establishing a communication link between the blade sensor and the main control system, low-frequency vibration signals of the blade can be collected, and an embedded fault diagnosis algorithm is used to realize real-time monitoring of blade impacts and safety protection, featuring high diagnostic coverage rate and a long mean time to failure.

Description

Technical Field

[0001] The present invention belongs to the technical field of wind turbine testing, and particularly relates to an online monitoring system, method, electronic device and medium for wind turbine blades. Background Art

[0002] In recent years, the development of ultra-large wind turbines has become a trend in the wind power generation industry. In order to improve the utilization rate of wind energy and increase power generation, the quality and length of wind turbine blades have been continuously increasing, and the influence of blades on the operation safety and stability of wind turbines is particularly obvious. In addition, due to surface or structural damage, icing, lightning damage of wind turbine blades, or excessive vibration and load exceeding the design limit caused by sudden changes in wind speed and direction, phenomena such as blade fracture or tower sweeping also occur from time to time. Therefore, it is particularly important for the online monitoring system of wind turbine blades to monitor the health status of the blades, timely detect and warn of potential faults, and improve the operation and maintenance efficiency and safety of the unit.

[0003] At present, the online monitoring system of wind turbine blades is used as an auxiliary control system and does not participate in unit control. When sudden faults such as blade fracture or tower sweeping occur in the unit, it is impossible to control the unit to quickly execute actions such as shutdown. Summary of the Invention

[0004] The purpose of the embodiments of the present invention is to provide an online monitoring system, method, electronic device and medium for wind turbine blades, which are used to completely or at least partially solve the technical problems in the above-mentioned prior art that it is impossible to timely detect potential blade faults and it is impossible to control the unit to shut down when a fault occurs.

[0005] In a first aspect, an embodiment of the present application provides an online monitoring system for wind turbine blades, including:

[0006] A low-frequency vibration sensor, installed at a position on the wind turbine blade at a preset distance from the blade root, for collecting vibration signals of the wind turbine blade;

[0007] A CANopen communication module, connected to the low-frequency vibration sensor, for receiving and transmitting the vibration signals of the wind turbine blade;

[0008] A CANopen master station terminal module, communicatively connected to the CANopen communication module, for transmitting the vibration signals of the wind turbine blade to the main control system;

[0009] The main control system processes and analyzes the blade vibration signals through an embedded fault diagnosis algorithm to realize real-time monitoring of the wind turbine blades.

[0010] Optionally, a 24VDC power lightning protection module is added to the CANopen communication module to protect the power supply from lightning strikes;

[0011] Add a CANopen lightning protection module to the CANopen master terminal module to protect the main control system from lightning strikes.

[0012] Optionally, the CANopen lightning protection module is used for 6V high-speed signal lines, with a voltage protection level less than 25V and a lightning impulse current of 5kA, for CAN network lightning protection;

[0013] The DC operating voltage of the 24VDC power supply lightning protection module is 24VDC, and the voltage protection level is less than or equal to 250V, for 24VDC power supply lightning protection.

[0014] Optionally, the type of the low-frequency vibration sensor is a capacitive biaxial accelerometer, which has a narrowband filter.

[0015] Optionally, the main control system processes and analyzes the blade vibration signals through an embedded fault diagnosis algorithm to realize real-time monitoring of the wind turbine blades, including:

[0016] Judge whether there is surface damage on the blade and its severity through the monitored blade surface damage index; judge whether there is structural damage on the blade and its severity through the blade structure damage index; judge whether there is imbalance on the blade and its severity through the blade imbalance index; judge whether there is icing on the blade through the blade icing monitoring index; characterize the first and second natural frequencies of the blade in the flapping and pitching directions through the natural frequency monitoring index; characterize the similarity degree of the operating states of the three blades through the data similarity index.

[0017] Optionally, the determination process of the blade surface damage index includes: two calculation methods of spectral centroid and differential discrete:

[0018] Use librosa.feature.spectral_centroid to calculate the spectral centroid of each frame of the vibration signal and find the median to obtain the first index value;

[0019] After using the differential discrete transformation of the time-domain waveform file, use the standard deviation to measure the degree of discreteness to obtain the second index value;

[0020] Combine the spectral centroid and differential discrete cases of the blade surface damage index, compare the monthly median with the preset threshold, determine the proportion of the first index value and the second index value exceeding the preset threshold, and take the maximum value of the three blades as the blade surface damage index;

[0021] The determination process of the blade structure damage index includes: measuring from the theoretical change rate and relative change rate of the pitching stiffness:

[0022] The ratio of the peak position of the first-order natural frequency of blade flapping obtained from the amplitude spectrum diagram to the actual first-order natural frequency of blade flapping is used to determine the theoretical change rate of the flapping stiffness;

[0023] The ratio of the peak position of the first-order natural frequency of flapping of a certain blade obtained from the amplitude spectrum diagram to the average value of the peak positions obtained from the remaining two blades is used to determine the relative change rate of the flapping stiffness;

[0024] Combining the two cases of the theoretical change rate and the relative change rate of the flapping stiffness, comparing the monthly median with the preset threshold, determining the proportion of the theoretical change rate and the relative change rate of the flapping stiffness that exceed the threshold, and taking the maximum value of the three blades as the blade structure damage index;

[0025] The determination process of the blade imbalance index includes:

[0026] Using the flapping signal of the time-domain waveform file, statistically calculating the RMS value near the 1p rotational frequency as the measurement value, and then calculating the range of the three blades to determine the flapping imbalance index;

[0027] Using the pitching signal of the time-domain waveform file, statistically calculating the RMS value near the 1p rotational frequency as the measurement value, and then calculating the range of the three blades to determine the pitching imbalance index;

[0028] Combining the flapping imbalance index and the pitching imbalance index, comparing the monthly median with the preset threshold, and taking the proportion of the flapping imbalance index and the pitching imbalance index that exceed the preset threshold as the blade imbalance index;

[0029] The determination process of the blade icing monitoring index includes:

[0030] Extracting the blade temperature value from the working condition file, where the external environmental condition for blade icing is the temperature value below degrees;

[0031] The determination process of the blade natural frequency monitoring index includes: the effective value of the first order of pitching and the peak position of the first order of pitching:

[0032] Statistically calculating the RMS effective value of the spectrum amplitude near the first-order natural frequency of pitching, and extracting the peak position near the first-order natural frequency of pitching. Comparing the monthly median with the threshold, and taking the proportion of the effective value of the first order of pitching and the peak position of the first order of pitching that exceed the preset threshold as the blade natural frequency monitoring index;

[0033] The determination process of the blade data similarity index includes:

[0034] Measuring the cosine similarity between every two of the three blades for the amplitude spectrum, and then statistically calculating the minimum similarity to determine the frequency-domain similarity;

[0035] Calculate the indicators in the time domain, then measure the cosine similarity between pairwise of the three blades for the sequence of time-domain indicators, and then count the minimum similarity to determine the time-domain similarity;

[0036] First, perform a first-order difference processing on the data of the time-domain waveform file, then count the time-domain indicator values, then measure the cosine similarity between pairwise of the three blades for the sequence of time-domain indicators, and then count the minimum similarity to determine the time-domain difference similarity;

[0037] Combined with three types of indicators: frequency domain, time domain, and time-domain difference, compare the monthly median with a preset threshold, and use the proportion of the three types of indicators (frequency domain, time domain, and time-domain difference) exceeding the preset threshold as the blade data similarity indicator.

[0038] In a second aspect, an embodiment of the present application further provides an online monitoring method for a wind turbine blade applied to the online monitoring system of the wind turbine blade, including:

[0039] Collect the vibration signals of the wind turbine blade through a low-frequency vibration sensor;

[0040] Use a preset fault diagnosis algorithm to process and analyze the blade vibration signals to achieve real-time monitoring of the wind turbine blade.

[0041] Optionally, using a preset fault diagnosis algorithm to process and analyze the blade vibration signals to achieve real-time monitoring of the wind turbine blade, including:

[0042] Based on the monitored blade surface damage indicators, judge whether there is surface damage on the blade and its severity; based on the blade structure damage indicators, judge whether there is structural damage on the blade and its severity; based on the blade imbalance indicators, judge whether there is imbalance on the blade and its severity; based on the blade icing monitoring indicators, judge whether there is icing on the blade; based on the natural frequency monitoring indicators, characterize the first-order and second-order natural frequencies of the blade in the flapping and pitching directions; based on the data similarity indicators, characterize the similarity degree of the operating states of the three blades; where

[0043] The determination process of the blade surface damage indicators includes: two calculation methods: spectral centroid and difference discrete:

[0044] Use librosa.feature.spectral_centroid to calculate the spectral centroid of each frame of the vibration signal and find the median to obtain the first indicator value;

[0045] After performing a difference discrete transformation on the time-domain waveform file, use the standard deviation to measure the degree of discreteness to obtain the second indicator value;

[0046] Combined with the spectral centroid and difference discrete cases of the blade surface damage index, compare the monthly median with the preset threshold, determine the proportions of the first index value and the second index value exceeding the preset threshold, and take the maximum value of the three blades as the blade surface damage index;

[0047] The determination process of the blade structure damage index includes: measuring from the theoretical change rate and relative change rate of the flapping stiffness:

[0048] Obtain the ratio of the peak position of the first-order natural frequency of blade flapping obtained from the amplitude spectrum diagram to the actual first-order natural frequency of blade flapping to determine the theoretical change rate of the flapping stiffness;

[0049] Obtain the ratio of the peak position of the first-order natural frequency of flapping of a certain blade obtained from the amplitude spectrum diagram to the average value of the peak positions obtained from the remaining two blades to determine the relative change rate of the flapping stiffness;

[0050] Combined with the two cases of the theoretical change rate and relative change rate of the flapping stiffness, compare the monthly median with the preset threshold, determine the proportions of the theoretical change rate and relative change rate of the flapping stiffness exceeding the threshold, and take the maximum value of the three blades as the blade structure damage index;

[0051] The determination process of the blade imbalance index includes:

[0052] Use the flapping signal of the time-domain waveform file, count the RMS value near the 1p rotational frequency as the measurement value, and then calculate the range of the three blades to determine the flapping imbalance index;

[0053] Use the pitching signal of the time-domain waveform file, count the RMS value near the 1p rotational frequency as the measurement value, and then calculate the range of the three blades to determine the pitching imbalance index;

[0054] Combined with the flapping imbalance index and the pitching imbalance index, compare the monthly median with the preset threshold, and take the proportions of the flapping imbalance index and the pitching imbalance index exceeding the preset threshold as the blade imbalance index;

[0055] The determination process of the blade icing monitoring index includes:

[0056] Extract the blade temperature value from the working condition file, where the external environmental condition for blade icing is the temperature value below degrees;

[0057] The determination process of the blade natural frequency monitoring index includes: the effective value of the first-order pitching and the peak position of the first-order pitching:

[0058] Perform RMS effective value statistics on the spectral amplitude near the first-order natural frequency of flapping, and extract the peak position near the first-order natural frequency of flapping. Compare the monthly median with the threshold, and use the ratio of the effective value of the first order of flapping and the peak position of the first order of flapping exceeding the preset threshold as the blade natural frequency monitoring index;

[0059] The determination process of the blade data similarity index includes:

[0060] Measure the cosine similarity between every two of the three blades for the amplitude spectrum, and then statistically determine the minimum similarity to obtain the frequency-domain similarity;

[0061] Perform time-domain index calculation, then measure the cosine similarity between every two of the three blades for the time-domain index sequence, and then statistically determine the minimum similarity to obtain the time-domain similarity;

[0062] First perform first-order difference processing on the data of the time-domain waveform file, then statistically determine the time-domain index value, then measure the cosine similarity between every two of the three blades for the time-domain index sequence, and then statistically determine the minimum similarity to obtain the time-domain differential similarity;

[0063] Combine the three types of indexes in the frequency domain, time domain, and time-domain difference. Compare the monthly median with the preset threshold, and use the ratio of the three types of indexes in the frequency domain, time domain, and time-domain difference exceeding the preset threshold as the blade data similarity index.

[0064] In a third aspect, an embodiment of the present application further provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, it implements the steps of the above-mentioned on-line monitoring method for the blades of a wind turbine generator set.

[0065] In a fourth aspect, an embodiment of the present application further provides a storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements the steps of the above-mentioned on-line monitoring method for the blades of a wind turbine generator set.

[0066] From the above technical solutions, it can be seen that the present invention has the following advantages:

[0067] In the on-line monitoring system, method, electronic device and medium for the blades of a wind turbine generator set provided by the present application, by establishing a communication link between the blade sensor and the PLC main control system, the low-frequency vibration signal of the blade can be collected, and a fault diagnosis algorithm is embedded to realize real-time monitoring of blade impact and safety protection, and has the characteristics of high diagnostic coverage rate and long mean time to failure. Description of the Drawings

[0068] To more clearly illustrate the technical solution of the present invention, the accompanying drawings required for description will be briefly introduced below. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other accompanying drawings can be obtained based on these drawings.

[0069] Figure 1 It is a schematic structural diagram of an on-line monitoring system for a wind turbine blade provided by an embodiment of the present invention;

[0070] Figure 2 It is an implementation flowchart of an on-line monitoring method for a wind turbine blade of an on-line monitoring system applied to a wind turbine blade provided by an embodiment of the present invention;

[0071] Figure 3 It is a schematic hardware structure diagram of an electronic device for each embodiment of the present invention. Detailed implementation manners

[0072] In the following detailed description, various embodiments of the present disclosure will be described more fully. The present disclosure can have various embodiments and adjustments and changes can be made therein. However, it should be understood that there is no intention to limit the various embodiments of the present disclosure to the specific embodiments disclosed herein, but the present disclosure should be understood to cover all adjustments, equivalents and / or alternative solutions falling within the spirit and scope of the various embodiments of the present disclosure.

[0073] In the following, the term "comprising" or "may comprise" that can be used in various embodiments of the present disclosure indicates the presence of the disclosed function or operation, and does not limit the addition of one or more functions or operations. In addition, as used in various embodiments of the present disclosure, the terms "comprising", "having" and their cognates are only intended to indicate a specific feature, number, step, operation or combination of the foregoing items, and should not be construed as first excluding the existence or addition of one or more other features, numbers, steps, operations or combinations of the foregoing items.

[0074] In various embodiments of the present disclosure, the expression "or" or "at least one of A or / and B" includes any combination or all combinations of the words listed simultaneously. For example, the expression "A or B" or "at least one of A or / and B" may include A, may include B, or may include both A and B.

[0075] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0076] Refer to Figure 1 The following is a schematic structural diagram of an on-line monitoring system for a wind turbine blade in a specific embodiment, including:

[0077] A low-frequency vibration sensor, installed at a position on the wind turbine blade at a preset distance from the blade root, for collecting the vibration signals of the wind turbine blade;

[0078] A CANopen communication module, connected to the low-frequency vibration sensor, for receiving and transmitting the vibration signals of the wind turbine blade;

[0079] A CANopen master station terminal module, communicatively connected to the CANopen communication module, for transmitting the vibration signals of the wind turbine blade to the main control system;

[0080] The main control system processes and analyzes the blade vibration signals through an embedded fault diagnosis algorithm to achieve real-time monitoring of the wind turbine blade.

[0081] Exemplarily, the low-frequency vibration sensor is installed at the 1 / 3 position from the blade root on the wind turbine blade, and collects the vibration signals in two directions of blade flap and flutter (the flap signal refers to the signal of bending vibration perpendicular to the rotation plane direction; the flutter signal refers to the signal of bending vibration in the rotation plane), and the main control system is a PLC main control system.

[0082] In a specific embodiment, a 24VDC power lightning protection module is added to the CANopen communication module to protect the power supply from lightning strikes; a CANopen lightning protection module is added to the CANopen master station terminal module to protect the main control system from lightning strikes. Among them, the CANopen lightning protection module is applied to a 6V high-speed signal line, with a voltage protection level less than 25V and a lightning impulse current of 5kA, for CAN network lightning protection; the DC operating voltage of the 24VDC power lightning protection module is 24VDC, and the voltage protection level is less than or equal to 250V, for 24VDC power lightning protection.

[0083] In this embodiment, the type of the low-frequency vibration sensor is a capacitive biaxial accelerometer, which has a narrowband filter, a frequency range of 0.05 - 1000Hz, detectors: safety shock detection (SSD), effective value, peak value, peak-to-peak value, and through a redundant safety relay, system faults can be output to the control system.

[0084] Preferably, the CANopen communication module has a 4-channel electrically isolated CAN interface, and the communication baud rate can be configured through the RS232 port: 5 Kbps - 1 Mbps, which is used to increase the number of nodes in the CAN network and change the topological structure of the CAN network wiring; the CANopen master station terminal module is a CANopen gateway for CANopen protocol conversion.

[0085] In a specific embodiment, Figure 1 it includes a pitch cabinet, a terminal block (2 in and 8 out), low-frequency vibration sensors (including low-frequency sensors A, B, and C with 120 Ω terminal resistors), a CANopen communication module (including a 24 VDC power lightning protection module), a slip ring communication cable, a CANopen master station terminal module (including a CANopen lightning protection module), and a PLC main control system. Among them, the red arrow represents the power supply 24V+, the blue arrow represents the power supply 24V-, the green arrow represents the CAN high level, the CAN low level, and the CAN ground.

[0086] Its working principle is as follows: Three low-frequency vibration sensors of the blades are installed at a position 1 / 3 from the blade root on the blades, and can collect vibration signals in two directions of blade flapping and pitching. The low-frequency vibration signals are transmitted to the CANopen communication module. To avoid lightning strikes, a 24 VDC power lightning protection module is added to protect the power supply. The signals pass through the slip ring communication cable and are transmitted to the PLC main control system through the CANopen master station terminal module. To avoid lightning strikes, a CANopen lightning protection module is added to protect the PLC main control system. The PLC main control system embeds fault diagnosis algorithms such as identifying blade surface or structural damage, blade imbalance, blade icing, natural frequency, and data similarity to realize real-time monitoring of blade impacts.

[0087] In a specific embodiment, the main control system processes and analyzes the blade vibration signals through the embedded fault diagnosis algorithm to realize real-time monitoring of the blades of the wind turbine, including:

[0088] By monitoring the blade surface damage index, judge whether there is blade surface damage and its severity; by the blade structure damage index, judge whether there is structural damage to the blade and its severity; by the blade imbalance index, judge whether there is imbalance of the blade and its severity; by the blade icing monitoring index, judge whether there is icing on the blade; by the natural frequency monitoring index, characterize the first and second natural frequencies in the flapping and pitching directions of the blade; by the data similarity index, characterize the similarity of the operating states of the three blades.

[0089] Specifically, the determination process of the blade surface damage index includes: two calculation methods of spectral centroid and differential separation:

[0090] The spectral centroid of each frame of the vibration signal is calculated using librosa.feature.spectral_centroid, and the median is obtained to get the first index value;

[0091] After discrete transformation by differential separation of the time-domain waveform file, the standard deviation is used to measure the degree of dispersion to obtain the second index value;

[0092] Combining the spectral centroid and differential separation of the blade surface damage index, the monthly median is compared with the preset threshold to determine the proportion of the first index value and the second index value exceeding the preset threshold, and the maximum value of the three blades is taken as the blade surface damage index;

[0093] The determination process of the blade structure damage index includes: measuring from the theoretical change rate and relative change rate of the flapping stiffness:

[0094] The ratio of the peak position of the first-order natural frequency of blade flapping obtained from the amplitude spectrum diagram to the actual first-order natural frequency of blade flapping is used to determine the theoretical change rate of the flapping stiffness;

[0095] The ratio of the peak position of the first-order natural frequency of blade flapping of a certain blade obtained from the amplitude spectrum diagram to the average value of the peak positions obtained from the remaining two blades is used to determine the relative change rate of the flapping stiffness;

[0096] Combining the theoretical change rate and relative change rate of the flapping stiffness, the monthly median is compared with the preset threshold to determine the proportion of the theoretical change rate and relative change rate of the flapping stiffness exceeding the threshold, and the maximum value of the three blades is taken as the blade structure damage index;

[0097] The determination process of the blade imbalance index includes:

[0098] The RMS value near the 1p rotational frequency is statistically calculated using the flapping signal of the time-domain waveform file as the measurement value, and then the range of the three blades is calculated to determine the flapping imbalance index;

[0099] The RMS value near the 1p rotational frequency is statistically calculated using the pitch signal of the time-domain waveform file as the measurement value, and then the range of the three blades is calculated to determine the pitch imbalance index;

[0100] Combining the flapping imbalance index and the pitch imbalance index, the monthly median is compared with the preset threshold, and the proportion of the flapping imbalance index and the pitch imbalance index exceeding the preset threshold is used as the blade imbalance index;

[0101] The determination process of the blade icing monitoring index includes:

[0102] Extract the blade temperature value from the operating condition file. Among them, the external environmental condition for blade icing is the temperature value below a certain degree;

[0103] The determination process of the blade natural frequency monitoring index includes: the effective value of the first flap order and the peak position of the first flap order:

[0104] Perform RMS effective value statistics on the spectral amplitude near the first flap natural frequency, and extract the peak position near the first flap natural frequency. Compare the monthly median with the threshold, and use the proportion of the effective value of the first flap order and the peak position of the first flap order exceeding the preset threshold as the blade natural frequency monitoring index;

[0105] The determination process of the blade data similarity index includes:

[0106] Measure the cosine similarity between every two of the three blades for the amplitude spectrum, and then statistically determine the minimum similarity to obtain the frequency domain similarity;

[0107] Perform index calculation in the time domain, then measure the cosine similarity between every two of the three blades for the sequence of time domain indexes, and then statistically determine the minimum similarity to obtain the time domain similarity;

[0108] First perform first-order difference processing on the data of the time domain waveform file, then statistically determine the time domain index value, then measure the cosine similarity between every two of the three blades for the sequence of time domain indexes, and then statistically determine the minimum similarity to obtain the time domain difference similarity;

[0109] Combined with three types of indexes in the frequency domain, time domain, and time domain difference, compare the monthly median with the preset threshold, and use the proportion of the three types of indexes in the frequency domain, time domain, and time domain difference exceeding the preset threshold as the blade data similarity index.

[0110] As Figure 2 shown, the following is an embodiment of the on-line monitoring method for wind turbine blades of the on-line monitoring system for wind turbine blades provided by the embodiments of the present disclosure. It belongs to the same inventive concept as the on-line monitoring system for wind turbine blades in the above embodiments. For the details not described in detail in the embodiment of the on-line monitoring method for wind turbine blades, reference can be made to the embodiments of the on-line monitoring system for wind turbine blades above.

[0111] The PLC main control system embeds a blade fault diagnosis algorithm to process and analyze the vibration data in step C. The characteristic indicators monitored by the system include the blade surface damage indicator, which is used to judge whether there is surface damage on the blade and its severity; the blade structural damage indicator, which is used to judge whether there is structural damage on the blade and its severity; the blade imbalance indicator, which is used to judge whether there is imbalance on the blade and its severity; the blade icing monitoring indicator, which is used to judge whether there is icing on the blade; the natural frequency monitoring indicator, which is used to characterize the first-order and second-order natural frequencies of the blade in the flapping and pitching directions; and the data similarity indicator, which is used to characterize the similarity degree of the operating states of the three blades. The PLC main control system can realize batch offline data import, display the fault indicators and intelligent diagnosis results, and the output data should at least include characteristic data. The output display graphs include trend graphs, time-domain graphs, and frequency-spectrum graphs, etc. The preprocessing functions include time screening, data deletion, and data selection. The operations include edge-frequency cursors, multiple-frequency cursors, zooming in, zooming out, and clicking on data to display the amplitude and time of various graphs.

[0112] Specifically, it includes the following execution steps:

[0113] Step 200: Collect the vibration signals of the wind turbine blades through low-frequency vibration sensors.

[0114] Exemplarily, taking a wind farm in Liaoning Province as an example, 24 3.XMW units are put into operation in the wind farm. Therefore, one unit is selected for the hanging test of the blade online real-time monitoring and safety protection system. The low-frequency vibration sensor is installed at the position 1 / 3 from the blade root on the wind turbine blade, and can collect the vibration signals in 2 directions of the blade flapping and pitching. The type of this sensor is a capacitive biaxial accelerometer, which has a narrow-band filter, with a frequency range: 0.05 - 1000Hz, and detectors: safety shock detection (SSD), root mean square value, peak value, peak-to-peak value. Through a redundant safety relay, the system fault is output to the control system; the CANopen communication module has 4-channel electrically isolated CAN interfaces, and the communication baud rate can be configured through the RS232 port: 5Kbps - 1Mbps; the CANopen master station terminal module is a CANopen gateway for CANopen protocol conversion; the PLC main control system can connect to and control the CANopen terminal device (low-frequency vibration sensor) through the CANopen master station terminal module; the CANopen lightning protection module is applied to 6V high-speed signal lines, with a voltage protection level < 25V and a lightning impulse current of 5kA; the DC working voltage of the 24VDC power supply lightning protection module is 24VDC, and the protection level ≤ 250V.

[0115] First, hang up and install the on-line real-time monitoring and safety protection system for wind turbine blades, including installing low-frequency vibration sensors at 1 / 3 of the distance from the blade root on each of the 3 blades. The sensor cables are installed in an "S" shape, routed at the blade cover plate position, and installed with mechanical interfaces and electrical interfaces, etc. Then, the sensors are powered by 24VDC from the pitch control cabinet, with a built-in 120Ω terminal resistor, and collect vibration signals in 2 directions of blade flap and lead-lag. Then, using the CANopen communication protocol, the signals are transmitted to the CANopen communication module, and through the slip ring communication cable, they are transmitted to the PLC main control system through the CANopen master station terminal module. The vibration data collected by the low-frequency vibration sensors mainly includes the original data of blades 1, 2, and 3, the frequency band data of blades 1, 2, and 3, and the status data of blades 1, 2, and 3.

[0116] Step 201: Use a preset fault diagnosis algorithm to process and analyze the blade vibration signals to achieve real-time monitoring of wind turbine blades.

[0117] Exemplarily, the PLC main control system embeds a blade fault diagnosis algorithm to process and analyze the vibration data. The characteristic indicators monitored by the system include the blade surface damage indicator, which is used to judge whether there is surface damage on the blade and its severity; the blade structure damage indicator, which is used to judge whether there is structural damage on the blade and its severity; the blade imbalance indicator, which is used to judge whether there is imbalance on the blade and its severity; the blade icing monitoring indicator, which is used to judge whether there is icing on the blade; the natural frequency monitoring indicator, which is used to characterize the first-order and second-order natural frequencies in the flap and lead-lag directions of the blade; and the data similarity indicator, which is used to characterize the similarity of the operating states of the three blades. The PLC main control system can realize batch offline data import, display the fault indicators and intelligent diagnosis results, and the output data should at least include characteristic data. The displayed graphics include trend charts, time-domain charts, and frequency-spectrum charts, etc. The preprocessing functions include time screening, data deletion, and data selection. The operations include sideband cursors, harmonic cursors, zoom in, zoom out, and point selection of data to display amplitude and time for various charts.

[0118] Specifically, when performing step 201, the following steps can be specifically executed:

[0119] Judge whether there is surface damage on the blade and its severity through the monitored blade surface damage indicator; judge whether there is structural damage on the blade and its severity through the blade structure damage indicator; judge whether there is imbalance on the blade and its severity through the blade imbalance indicator; judge whether there is icing on the blade through the blade icing monitoring indicator; characterize the first-order and second-order natural frequencies in the flap and lead-lag directions of the blade through the natural frequency monitoring indicator; characterize the similarity of the operating states of the three blades through the data similarity indicator; where

[0120] The determination process of the blade surface damage index includes two calculation methods: spectral centroid and differential discrete:

[0121] Use librosa.feature.spectral_centroid to calculate the spectral centroid of each frame of the vibration signal and find the median to obtain the first index value;

[0122] After using the differential discrete transformation of the time-domain waveform file, use the standard deviation to measure the degree of dispersion to obtain the second index value;

[0123] Combining the two cases of spectral centroid and differential discrete of the blade surface damage index, compare the monthly median with the preset threshold, determine the proportion of the first index value and the second index value exceeding the preset threshold, and take the maximum value of the three blades as the blade surface damage index;

[0124] The determination process of the blade structure damage index includes: measuring from the theoretical change rate and relative change rate of the flapping stiffness:

[0125] Through the amplitude spectrum diagram, obtain the ratio of the peak position of the first natural frequency of the blade flapping to the actual first natural frequency of the blade flapping to determine the theoretical change rate of the flapping stiffness;

[0126] Through the amplitude spectrum diagram, obtain the ratio of the peak position of the first natural frequency of the flapping of a certain blade to the average value of the peak positions obtained by the remaining two blades to determine the relative change rate of the flapping stiffness;

[0127] Combining the two cases of the theoretical change rate and relative change rate of the flapping stiffness, compare the monthly median with the preset threshold, determine the proportion of the theoretical change rate and relative change rate of the flapping stiffness exceeding the threshold, and take the maximum value of the three blades as the blade structure damage index;

[0128] The determination process of the blade imbalance index includes:

[0129] Use the flapping signal of the time-domain waveform file, count the RMS value near the 1p rotational frequency as the measurement value, and then calculate the range of the three blades to determine the flapping imbalance index;

[0130] Use the pitching signal of the time-domain waveform file, count the RMS value near the 1p rotational frequency as the measurement value, and then calculate the range of the three blades to determine the pitching imbalance index;

[0131] Combining the flapping imbalance index and the pitching imbalance index, compare the monthly median with the preset threshold, and use the proportion of the flapping imbalance index and the pitching imbalance index exceeding the preset threshold as the blade imbalance index;

[0132] The determination process of the blade icing monitoring index includes:

[0133] Extract the blade temperature value from the operating condition file. Among them, the external environmental condition for blade icing is the temperature value below degrees;

[0134] The determination process of the blade natural frequency monitoring index includes: the effective value of the first flap order and the peak position of the first flap order:

[0135] Perform RMS effective value statistics on the spectral amplitude near the first flap natural frequency, and extract the peak position near the first flap natural frequency. Compare the monthly median with the threshold, and use the proportion of the effective value of the first flap order and the peak position of the first flap order exceeding the preset threshold as the blade natural frequency monitoring index;

[0136] The determination process of the blade data similarity index includes:

[0137] Measure the cosine similarity between two of the three blades for the amplitude spectrum, and then statistically determine the minimum similarity to obtain the frequency domain similarity;

[0138] Perform index calculation in the time domain, then measure the cosine similarity between two of the three blades for the sequence of time domain indexes, and then statistically determine the minimum similarity to obtain the time domain similarity;

[0139] First perform first-order difference processing on the data of the time domain waveform file, then statistically determine the time domain index value, then measure the cosine similarity between two of the three blades for the sequence of time domain indexes, and then statistically determine the minimum similarity to obtain the time domain difference similarity;

[0140] Combined with three types of indexes in the frequency domain, time domain, and time domain difference, compare the monthly median with the preset threshold, and use the proportion of the three types of indexes in the frequency domain, time domain, and time domain difference exceeding the preset threshold as the blade data similarity index.

[0141] By establishing a communication link between the blade sensor and the PLC main control system, the low-frequency vibration signal of the blade can be collected, and fault diagnosis algorithms for identifying blade surface or structural damage, blade imbalance, blade icing, natural frequency, data similarity, etc. can be embedded. Its purpose is to realize real-time monitoring and safety protection of blade impact, and has characteristics such as high diagnostic coverage rate and long mean time to failure.

[0142] Figure 3 It is a schematic diagram of the hardware structure of an electronic device for implementing each embodiment of the present invention.

Claims

1. An online monitoring system for wind turbine blades, characterized in that: include: A low-frequency vibration sensor is installed at a preset distance between the blade of the wind turbine and the blade root to collect vibration signals of the blade of the wind turbine; A CANopen communication module, connected to the low-frequency vibration sensor, for receiving and transmitting vibration signals of the wind turbine blades; A CANopen master terminal module, which is communicatively connected to the CANopen communication module and is used to transmit the vibration signal of the wind turbine blade to the main control system; The main control system processes and analyzes blade vibration signals through an embedded fault diagnosis algorithm to achieve real-time monitoring of wind turbine blades; Wherein, the type of the low-frequency vibration sensor is a capacitive dual-axis accelerometer with a narrow-band filter; The main control system processes and analyzes blade vibration signals through an embedded fault diagnosis algorithm to achieve real-time monitoring of wind turbine blades, including: Through the monitored blade surface damage index, it is judged whether the blade has surface damage and its severity; through the blade structure damage index, it is judged whether the blade has structural damage and its severity; through the blade imbalance index, it is judged whether the blade is unbalanced and its severity; through the blade icing monitoring index, it is judged whether the blade has ice; through the natural frequency monitoring index, it characterizes the first-order and second-order natural frequencies of the blade in the flapping and swinging directions; through the data similarity index, it characterizes the similarity of the operating conditions of the three blades; The process of determining the blade structure damage index includes: measuring from the theoretical change rate and relative change rate of the swing stiffness: The theoretical change rate of the shimmy stiffness is determined by the ratio of the peak position of the blade shimmy natural first-order frequency obtained through the amplitude spectrum diagram and the actual shimmy natural first-order frequency of the blade; The relative change rate of the shimmy stiffness is determined by the ratio of the peak position of the first-order frequency of the shimmy inherent of a blade obtained through the amplitude spectrum diagram to the average of the peak positions obtained by the remaining two blades; Combining the theoretical change rate and relative change rate of the shimmy stiffness, the monthly median is compared with the preset threshold to determine the proportion of the theoretical change rate and relative change rate of the shimmy stiffness exceeding the threshold, and the maximum value of the three blades is taken as the blade structure damage index; The process of determining the blade imbalance index includes: The flapping signal of the time domain waveform file is used to calculate the RMS value near the 1p rotation frequency as the measurement value, and then the range of the three blades is calculated to determine the flapping imbalance index; The shimmy signal of the time domain waveform file is used to calculate the RMS value near the 1p rotation frequency as the measurement value, and then the range of the three blades is calculated to determine the shimmy imbalance index; Combine the flapping imbalance index and the shimmy imbalance index, compare the monthly median with the preset threshold, and take the proportion of the flapping imbalance index and the shimmy imbalance index exceeding the preset threshold as the blade imbalance index; The process of determining the leaf data similarity index includes: The cosine similarity between the three blades of the amplitude spectrum is measured, and then the minimum similarity is statistically calculated to determine the frequency domain similarity; Calculate the time domain index, measure the cosine similarity between the three leaves of the time domain index sequence, and then calculate the minimum similarity to determine the time domain similarity; The data of the time domain waveform file is first processed by first-order difference, and then the time domain index value is counted, and then the cosine similarity between the three leaves of the sequence of the time domain index is measured, and then the minimum similarity is counted to determine the time domain difference similarity; The three types of indicators, namely frequency domain, time domain and time domain difference, are combined. The monthly median is compared with the preset threshold, and the proportion of the three types of indicators, namely frequency domain, time domain and time domain difference, exceeding the preset threshold is used as the leaf data similarity index.

2. The online monitoring system for wind turbine blades according to claim 1, characterized in that: A 24VDC power supply lightning protection module is added to the CANopen communication module to protect the power supply from being struck by lightning; A CANopen lightning protection module is added to the CANopen master terminal module to protect the main control system from being struck by lightning.

3. The online monitoring system for wind turbine blades according to claim 2, characterized in that: CANopen lightning protection module is used for 6V high-speed signal line, with voltage protection level less than 25V and lightning impulse current of 5kA, and is used for lightning protection of CAN network; The DC working voltage of the 24VDC power supply lightning protection module is 24VDC, and the voltage protection level is less than or equal to 250V. It is used for 24VDC power supply lightning protection.

4. The online monitoring system for wind turbine blades according to claim 1, characterized in that: The process of determining the blade surface damage index includes two calculation methods: spectrum centroid and differential dispersion: Use librosa.feature.spectral_centroid to calculate the spectral centroid of each frame of the vibration signal and find the median to obtain the first index value; After using the time domain waveform file for differential discrete transformation, the standard deviation is used to measure the discrete degree to obtain the second index value; Combining the spectrum centroid and differential dispersion of the blade surface damage index, the monthly median is compared with the preset threshold to determine the ratio of the first index value and the second index value exceeding the preset threshold, and the maximum value of the three leaves is taken as the blade surface damage index; The process of determining blade icing monitoring indicators includes: Extract the blade temperature value in the working condition file, where the external environmental condition for blade icing is a temperature value below 0°C; The process of determining the blade natural frequency monitoring index includes: the effective value of the first-order shimmy and the peak position of the first-order shimmy: The RMS effective value of the spectrum amplitude near the first-order natural frequency of the swing vibration is statistically calculated, and the peak position near the first-order natural frequency of the swing vibration is extracted. The monthly median is compared with the threshold, and the ratio of the effective value of the first-order swing vibration and the peak position of the first-order swing vibration that exceeds the preset threshold is used as the blade natural frequency monitoring indicator.

5. An online monitoring method for wind turbine blades applied to the online monitoring system for wind turbine blades according to any one of claims 1 to 4, characterized in that: include: The vibration signals of the wind turbine blades are collected through low-frequency vibration sensors; Use the preset fault diagnosis algorithm to process and analyze the blade vibration signal to achieve real-time monitoring of wind turbine blades; Among them, the blade vibration signal is processed and analyzed using a preset fault diagnosis algorithm to achieve real-time monitoring of wind turbine blades, including: Through the monitored blade surface damage index, it is judged whether the blade has surface damage and its severity; through the blade structure damage index, it is judged whether the blade has structural damage and its severity; through the blade imbalance index, it is judged whether the blade is unbalanced and its severity; through the blade icing monitoring index, it is judged whether the blade has ice; through the natural frequency monitoring index, the first-order and second-order natural frequencies of the blade in the flapping and swinging directions are characterized; through the data similarity index, the similarity of the operating conditions of the three blades is characterized; among them, The process of determining the blade structure damage index includes: measuring from the theoretical change rate and relative change rate of the swing stiffness: The theoretical change rate of the shimmy stiffness is determined by the ratio of the peak position of the blade shimmy natural first-order frequency obtained through the amplitude spectrum diagram and the actual shimmy natural first-order frequency of the blade; The relative change rate of the shimmy stiffness is determined by the ratio of the peak position of the first-order frequency of the shimmy inherent of a blade obtained through the amplitude spectrum diagram to the average of the peak positions obtained by the remaining two blades; Combining the theoretical change rate and relative change rate of the shimmy stiffness, the monthly median is compared with the preset threshold to determine the proportion of the theoretical change rate and relative change rate of the shimmy stiffness exceeding the threshold, and the maximum value of the three blades is taken as the blade structure damage index; The process of determining the blade imbalance index includes: The flapping signal of the time domain waveform file is used to calculate the RMS value near the 1p rotation frequency as the measurement value, and then the range of the three blades is calculated to determine the flapping imbalance index; The shimmy signal of the time domain waveform file is used to calculate the RMS value near the 1p rotation frequency as the measurement value, and then the range of the three blades is calculated to determine the shimmy imbalance index; Combine the flapping imbalance index and the shimmy imbalance index, compare the monthly median with the preset threshold, and take the proportion of the flapping imbalance index and the shimmy imbalance index exceeding the preset threshold as the blade imbalance index; The process of determining the leaf data similarity index includes: The cosine similarity between the three blades of the amplitude spectrum is measured, and then the minimum similarity is statistically calculated to determine the frequency domain similarity; Calculate the time domain index, measure the cosine similarity between the three leaves of the time domain index sequence, and then calculate the minimum similarity to determine the time domain similarity; The data of the time domain waveform file is first processed by first-order difference, and then the time domain index value is counted, and then the cosine similarity between the three leaves of the sequence of the time domain index is measured, and then the minimum similarity is counted to determine the time domain difference similarity; The three types of indicators, namely frequency domain, time domain and time domain difference, are combined. The monthly median is compared with the preset threshold, and the proportion of the three types of indicators, namely frequency domain, time domain and time domain difference, exceeding the preset threshold is used as the leaf data similarity index.

6. The online monitoring method for wind turbine blades according to claim 5, characterized in that: The process of determining the blade surface damage index includes two calculation methods: spectrum centroid and differential dispersion: Use librosa.feature.spectral_centroid to calculate the spectral centroid of each frame of the vibration signal and find the median to obtain the first index value; After using the time domain waveform file for differential discrete transformation, the standard deviation is used to measure the discrete degree to obtain the second index value; Combining the spectrum centroid and differential dispersion of the blade surface damage index, the monthly median is compared with the preset threshold to determine the ratio of the first index value and the second index value exceeding the preset threshold, and the maximum value of the three leaves is taken as the blade surface damage index; The process of determining blade icing monitoring indicators includes: Extract the blade temperature value in the working condition file, where the external environmental condition for blade icing is a temperature value below 0°C; The process of determining the blade natural frequency monitoring index includes: the effective value of the first-order shimmy and the peak position of the first-order shimmy: The RMS effective value of the spectrum amplitude near the first-order natural frequency of the swing vibration is statistically calculated, and the peak position near the first-order natural frequency of the swing vibration is extracted. The monthly median is compared with the threshold, and the ratio of the effective value of the first-order swing vibration and the peak position of the first-order swing vibration that exceeds the preset threshold is used as the blade natural frequency monitoring indicator.

7. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the steps of the online monitoring method for wind turbine blades as claimed in claim 5 are implemented.

8. A storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the online monitoring method for wind turbine blades as claimed in claim 5 are implemented.

Images