Wind driven generator motor fault on-line detection method based on current perception

Abstract

The invention belongs to the technical field of wind power generation, and discloses a wind driven generator motor fault online detection method based on current perception. According to the method, motor current data is acquired in real time based on a pre-established current acquisition system, a cloud server in the current acquisition system firstly receives current data of a corresponding motor uploaded by each current transformer, and performs data cleaning on the current data of each motor; and then performing threshold comparison on the motor current data after the starting current of the motor is filtered out, performing comparison on transverse current of multiple motors of the same type and comparison on longitudinal historical current data of a single motor based on a KNN clustering algorithm, and judging whether the running state of the motor is normal or not in a grading manner. According to the wind driven generator motor fault on-line detection method, the defect that a vibration monitoring method can only analyze mechanical faults of the motor is overcome, and the accuracy of motor fault judgment is greatly improved through multi-dimensional early warning.

Description

Technical Field

[0001] The invention belongs to the field of wind power generation and relates to an online detection method for motor faults of a wind power generator based on current sensing. Background Art

[0002] At present, there are two main types of motor failures in wind turbines. The first is yaw motor failure. Wind turbines are mostly installed in coastal and hilly areas, where the turbulence intensity is relatively large. The unit transmission system needs to withstand more fatigue loads, and the transmission function of the wind turbine unit yaw system mainly relies on the yaw motor and yaw reducer transmission. The unit yaw braking function relies on the yaw motor electromagnetic brake braking. The yaw motor failure seriously affects the motor's yaw direction adjustment and restricts the wind power generation efficiency; the second is the heat dissipation motor failure. There are many types of ventilation and heat dissipation motors in the heat dissipation system inside the wind turbine cabin. The failure of the heat dissipation motor affects the mechanical and electrical systems of the entire wind turbine. The failure of any heat dissipation motor will affect the normal operation of the wind turbine, and even seriously affect the service life of the wind turbine system.

[0003] At present, the main problems that cause the yaw motor and heat dissipation motor to fail are motor bearing damage and jamming, and short circuits between phases or turns of the motor internal winding. At present, the fault monitoring of yaw motors and heat dissipation motors is mainly carried out through vibration monitoring methods, such as Figure 2 As shown in the figure, this method installs vibration sensors in the horizontal and vertical directions of the yaw motor and the heat dissipation motor to measure their vibration peak value, effective value, frequency and other parameters, and judges the running state and condition of the motor through the vibration frequency and other data when the motor is running, thereby providing data reference for motor fault monitoring. However, vibration monitoring has the following problems:

[0004] First, vibration monitoring can only determine whether there are mechanical problems with the motor, but cannot determine whether there are circuit problems with the motor; the core monitoring goal of the vibration monitoring solution is to analyze mechanical problems of motor bearings, gears, etc. through vibration, but mechanical problems are only one of the manifestations of motor failures. The vibration monitoring method cannot analyze circuit problems, so it has certain limitations. Second, there are accuracy problems with vibration monitoring. The vibration sensor is affected by the on-site installation environment and the vibration interference of other equipment, which can easily cause misjudgment of the motor's operating status. In addition, the frequency changes in the early stage of motor failure are slight, which is difficult for the vibration sensor to capture. Third, on-site construction is difficult and deployment is difficult. The yaw motor and heat dissipation motor have fixed positions in the wind turbine generator set, and most motors have small installation spaces. The small cabin space also makes wiring and routing very difficult, making it difficult or impossible to install vibration sensors. The construction difficulty and cost are relatively high.

[0005] Based on this, there is an urgent need for a wind turbine motor fault detection solution to ensure the stable operation of the wind turbine equipment. Summary of the invention

[0006] The purpose of the present invention is to propose an online detection method for wind turbine motor faults based on current sensing. The method is based on a pre-built current acquisition system, and remotely acquires the current data of each motor in the wind turbine in real time. The current data of each motor is cleaned, and then threshold comparison, lateral current comparison of multiple motors of the same type, and longitudinal historical current data comparison of a single motor are performed to determine whether the motor operating status is normal in a graded manner.

[0007] In order to achieve the above object, the present invention adopts the following technical scheme:

[0008] The online detection method of wind turbine motor fault based on current sensing is implemented based on a pre-built current acquisition system; the current acquisition system includes a cloud server and multiple current transformers;

[0009] Each current transformer collects the current data of one phase of the three-phase current of any motor in the wind turbine generator, and the collected current data is uploaded to the cloud server via the intelligent gateway;

[0010] The wind turbine generator motor fault online detection method based on current sensing comprises the following steps:

[0011] Step 1. The cloud server receives the current data of the corresponding motor uploaded by each current transformer in the current acquisition system, and performs data cleaning on the current data to filter out the starting current of the corresponding motor;

[0012] Step 2. Perform a rough threshold judgment on the motor current data after the data cleaning in step 1;

[0013] When the motor current value collected by the current transformer exceeds the preset threshold, it is directly determined that the motor is faulty; when the motor current value collected by the current transformer does not exceed the preset threshold, go to step 3;

[0014] Step 3. Compare the lateral currents of the same type and multiple motors based on KNN clustering for the motor currents that do not exceed the threshold;

[0015] For motor currents with relatively small data differences, go to step 4 to perform longitudinal historical current data comparison; for motor currents with relatively large data differences, compare them with the motor rated current to determine whether the motor current is within the preset safety range;

[0016] If the motor current value is within the preset safety range, the motor current is further transferred to step 4 for longitudinal historical current data comparison. If it is not within the preset safety range, it is directly determined that the motor is abnormal.

[0017] Step 4. Further compare the motor current value with its own historical normal value of motor current;

[0018] The average value of the historical normal value of the motor current is the average value of the historical current data of the motor under normal operation;

[0019] If the difference between the motor current value and the average of the motor current's historical normal values ​​is within the preset range, the motor is deemed to be operating normally; otherwise, the difference between the motor current value and its historical normal values ​​is further determined.

[0020] If the difference trend increases gradually over time, it indicates that the circuit or internal mechanical problem is gradually worsening. If the difference trend becomes stable over time, it is judged that the circuit or internal mechanical problem is caused by dust or circuit aging.

[0021] In addition, based on the online detection method of wind turbine motor fault based on current sensing, the present invention also proposes a corresponding online detection system of wind turbine motor fault based on current sensing, and its technical solution is as follows:

[0022] A wind turbine motor fault online detection system based on current sensing, comprising:

[0023] A data cleaning module is used to clean the current data of the corresponding motor uploaded by each current transformer in the current acquisition system received by the cloud server to filter out the starting current of the motor;

[0024] A threshold judgment module is used to make a rough threshold judgment on the motor current data after data cleaning;

[0025] When the motor current value collected by the current transformer exceeds the preset threshold, it is determined that the motor is faulty; when the motor current value collected by the current transformer does not exceed the preset threshold, it switches to the lateral current comparison module of multiple motors of the same type;

[0026] The same type of multi-motor transverse current comparison module is used to further compare the same type and multi-motor transverse currents based on the KNN clustering algorithm for motor currents that do not exceed the threshold;

[0027] The motor current with relatively small data difference is transferred to the longitudinal historical current data comparison module for judgment; the motor current with relatively large data difference is compared with the rated current value to determine whether it is within the preset safety range;

[0028] If the motor current is within the preset safety range, it will be transferred to the longitudinal historical current data comparison module for further judgment. If the motor current is not within the preset safety range, it will be directly judged as a motor abnormality;

[0029] and a longitudinal historical current data comparison module for comparing the motor current value with its own historical normal value of the motor current;

[0030] The average value of the historical normal value of the motor current is the average value of the historical current data of the motor under normal operation;

[0031] If the difference between the motor current value and the average of the motor current's historical normal values ​​is within the preset range, the motor is deemed to be operating normally; otherwise, the difference between the motor current value and its historical normal values ​​is further determined.

[0032] The specific situation of the motor abnormality can be further distinguished based on whether the difference development trend changes smoothly over time.

[0033] In addition, based on the above-mentioned current sensing-based online detection method for wind turbine motor faults, the present invention also proposes a computer device, which includes a memory and one or more processors; the memory stores executable code, and when the processor executes the executable code, it is used to implement the steps of the above-mentioned current sensing-based online detection method for wind turbine motor faults.

[0034] In addition, based on the above-mentioned current sensing-based online detection method for wind turbine motor faults, the present invention also proposes a computer-readable storage medium on which a program is stored; when the program is executed by a processor, it is used to implement the steps of the above-mentioned current sensing-based online detection method for wind turbine motor faults.

[0035] The present invention has the following advantages:

[0036] As described above, the present invention relates to an online detection method for wind turbine motor faults based on current sensing. This method solves the drawback that traditional vibration monitoring methods can only analyze motor mechanical faults. The motor current data can quickly locate whether it is a motor line problem or a motor mechanical fault problem. For example, when the A, B, and C three-phase current data of the motor operation all exceed the preset threshold value, it is confirmed that the motor has a major mechanical fault that causes an overall increase in the motor current; when the current value of one phase of the A, B, and C three-phase current of the motor is 0 or the current value of one phase increases, it is determined that there is a problem with the motor line. When the current of one phase is 0, the corresponding current is missing phase. When the current value of one phase increases, the corresponding line terminal is loose, resulting in an increase in resistance. After the motor current value is compared with the horizontal current of the same type and multiple motors and the longitudinal historical current data of a single motor, the development and changes of the motor's operating health status can be finely judged. For example, if the current of a certain motor does not exceed the preset threshold, it means that the motor is operating normally at the current stage. The motor current enters the horizontal current comparison stage of the same type and multiple motors. If the motor current value deviates from the current of the same type of motor for horizontal comparison, it is judged that a small mechanical fault has occurred inside the motor or a phase line is loose. The reaction to the change of current is more intuitive than the reaction to the change of vibration. The method of the present invention adopts multi-dimensional early warning. After the current data of each motor is filtered out of the motor's starting current through data cleaning, the threshold comparison, the horizontal current comparison of multiple motors of the same type, and the longitudinal historical current data comparison of a single motor are performed. The hierarchical judgment method combining rough judgment and refined judgment greatly improves the accuracy of motor fault judgment. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 It is a structural diagram of a method for online detection of motor faults of a wind turbine generator based on current sensing in an embodiment of the present invention;

[0038] Figure 2 is a schematic diagram of a vibration detection method;

[0039] Figure 3 A schematic diagram of a current acquisition system according to an embodiment of the present invention;

[0040] Figure 4 A schematic diagram of a current acquisition main control box and a current transformer in a current acquisition system in an embodiment of the present invention;

[0041] Figure 5 A schematic diagram of a current acquisition main control box in an embodiment of the present invention;

[0042] Figure 6 Schematic diagram of the motor starting current in an embodiment of the present invention;

[0043] Figure 7 A flowchart of data cleaning in an embodiment of the present invention;

[0044] Figure 8 This is a flow chart of a rough threshold determination in an embodiment of the present invention;

[0045] Fig. 9 A flowchart of obtaining critical values ​​of motor rated current and motor fault current by binary division in an embodiment of the present invention;

[0046] Fig.10 This is a flow chart of the comparison of transverse currents of multiple motors of the same type in an embodiment of the present invention;

[0047] Fig.11 A flowchart for setting the K value in an embodiment of the present invention;

[0048] Fig.12 A flowchart of longitudinal historical current data comparison in an embodiment of the present invention;

[0049] Fig.13 A flow chart of horizontally comparing and clustering multiple groups of motors to determine motor line problems and potential motor failures in an embodiment of the present invention;

[0050] Fig.14 This is a flow chart of judging motor abnormality by comparing longitudinal historical current data in an embodiment of the present invention. DETAILED DESCRIPTION

[0051] The present invention is further described in detail below with reference to the accompanying drawings and specific embodiments:

[0052] Example 1

[0053] Before introducing the online detection method for wind turbine motor fault based on current sensing in the present invention, a system is pre-built as follows: Figure 3 The current acquisition system shown in the figure includes the underlying sensing layer equipment and the cloud server.

[0054] Specifically, the bottom sensing layer equipment includes a current transmitter and multiple current transformers.

[0055] There are two types of motors in a wind turbine, including yaw motors and heat dissipation motors. There are multiple yaw motors and heat dissipation motors, and each current transformer is used to collect one phase current data of the three currents of a motor in the wind turbine.

[0056] For example, if a motor has three-phase currents A, B, and C, each motor corresponds to three current transformers, wherein each current transformer collects the A-phase, B-phase, and C-phase currents of the motor respectively.

[0057] Similarly, assuming that there are M yaw motors and heat dissipation motors in total, 3M current transformers are needed to collect current data. Here, M is a natural number greater than 0, which is related to the actual number of yaw motors and heat dissipation motors.

[0058] Each current transformer converts the collected current data from analog data to digital data through a current transmitter, and uploads the collected motor current data to the cloud server via wireless transmission.

[0059] Each wind turbine corresponds to an intelligent gateway and a current transmitter. One current transmitter is connected to multiple current transformers. The current data collected by each current transformer is transmitted to the same cloud server via the corresponding intelligent gateway.

[0060] The cloud server is used to parse the uploaded current data to determine whether the motor is operating normally by comparing the current threshold of each motor, comparing the currents of multiple motors of the same type horizontally, and comparing the historical current data of a single motor vertically.

[0061] The deployment of the current acquisition main control box and current transformer in the current acquisition system is as follows: Figure 4 shown.

[0062] The current transformer is installed at the fan distribution cabinet, and the current acquisition main control box is installed near the fan distribution cabinet. The current acquisition main control box is responsible for collecting data from the current transformer. The leads of the current transformer are led out from the fan distribution cabinet and connected to the current acquisition main control box to complete data collection. The collected data will be converted from analog to digital through the current transformer.

[0063] Specifically, the current acquisition main control box structure in this embodiment is as follows: Figure 5 The current acquisition main control box includes an air switch, a 12V power supply, an intelligent gateway, a current transmitter and a terminal block.

[0064] The power cord of the current acquisition main control box is connected to the 12V power supply, and the 12V power supply is used to power the smart gateway and current transmitter. The analog quantity collected by the current transformer is transmitted to the current transmitter through the terminal block.

[0065] The current transmitter converts the analog value of the current into standard modbus485 data and sends it to the intelligent gateway. The intelligent gateway uploads the motor current data to the cloud server via wireless transmission.

[0066] In this embodiment, the current transformer preferably adopts the AKH-0.66W-7 series current transformer, which has the advantages of small size, high precision, strong load capacity, etc., and can output small current signals such as 5mA-20mA.

[0067] In addition to its high data acquisition accuracy, the current transformer is also easy to install on site. The fan distribution cabinet supplies power to each motor, and the current transformer is installed in the fan distribution cabinet where the motor is connected.

[0068] The current transformers are centrally installed in the fan distribution cabinet, which has high space utilization, is easy to fix and simple to deploy. It avoids the problems of difficult installation and wiring of vibration sensors and insufficient installation space, reduces the difficulty of installation work and improves installation efficiency.

[0069] like Figure 1 As shown, after clearly introducing the pre-built current acquisition system, the online detection method for wind turbine motor fault based on current sensing in this embodiment is described in detail below.

[0070] The online detection method for wind turbine motor fault based on current sensing in this embodiment includes the following steps:

[0071] Step 1. The cloud server receives the current data of the corresponding motor uploaded by each current transformer in the current acquisition system, and performs data cleaning on the current data to filter out the starting current of the motor.

[0072] The current data uploaded by the smart gateway must first be cleaned by the cloud server to filter out the motor's starting current. The purpose of data cleaning is to ensure the authenticity and reliability of the current data.

[0073] like Figure 6 As shown, the horizontal axis t represents time, the vertical axis I represents the current value, I e Indicates the rated current. Since the motor is in a stationary state at the moment of power-on, the rotor in the motor is stationary. At this time, the rotor speed is 0, and the synchronous rotating magnetic field cuts the rotor winding at the maximum cutting speed, so that the rotor winding is induced and reaches the highest electromotive force, generating a large current in the rotor winding. This current will offset the magnetic flux of the stator magnetic field in the motor. In order to maintain the original magnetic flux that is compatible with the power supply voltage, the stator winding will automatically increase the current. Because the rotor current is very large at this time, the stator current in the motor will also increase greatly, up to 5 to 7 times the rated current, so it is necessary to clean and filter the motor starting current data to ensure that the motor current data of normal operation is obtained, that is, the motor starting current motor current data is filtered out after the data is cleaned.

[0074] Since the starting current of the motor is maintained for a very short time, usually not exceeding 0.5s, the current data is filtered by the online detection method for wind turbine motor fault based on current sensing in this embodiment, as shown in FIG. Figure 7 shown.

[0075] The motor current data collected by the current acquisition system is judged. When it is identified that the motor current starts to change from 0A to non-0A, the data within 2s of the current motor current is defined as the starting current. This part of the data, i.e. the defined starting current data, is filtered, and the data of the current motor current after 2s is used for the rough threshold judgment of step 2.

[0076] Step 2. Perform a rough threshold judgment on the motor current data after the data is cleaned in step 1.

[0077] When the motor current value collected by the current transformer exceeds the preset threshold, it is directly determined that the motor is faulty; when the motor current value collected by the current transformer does not exceed the preset threshold, go to step 3.

[0078] The expression of threshold judgment is the most direct. When the motor has bearing damage, jamming, or internal windings, the resistance of the motor increases, causing the current to change. The threshold judgment method can intuitively and effectively screen out the faulty motor. Specifically, when the motor has bearing damage, jamming, or internal windings, the resistance of the motor increases, causing the current to change. At this time, the values ​​of the three-phase currents A, B, and C all increase. Or, for example, if the motor is missing a phase, causing the current of one phase to be 0, but the currents of the other two phases increase, then it is determined that the motor has a phase loss problem.

[0079] The process of rough threshold judgment is as follows: Figure 8 shown.

[0080] First, the threshold parameters of the motor current data are preset. The motor current data collected by the current acquisition system is compared with the preset threshold parameters after data cleaning. When the motor current data exceeds this threshold, it will directly trigger an early warning push to prompt motor failure.

[0081] Whether the threshold rough judgment can play a warning role depends on the setting of the warning threshold, such as Fig. 9 shown.

[0082] The method of the present invention combines the rated current of the motor itself and the historical data of the motor current collected by the cloud server, takes out the current value in the time period when the equipment fails, that is, the faulty motor current, and uses the dichotomy method to obtain the critical value of the motor rated current and the motor fault current, and sets the critical value as the current warning threshold, that is, the preset threshold in the threshold rough judgment.

[0083] The specific process of obtaining the critical values ​​of the motor rated current and the motor fault current by using the dichotomy method is as follows:

[0084] Based on the current value collected from the current faulty motor as the upper limit of the current warning threshold, the rated current is used as the reference lower limit for setting the current warning threshold, and the middle value between the upper limit and the rated current is determined by binary search. The middle value is used as the upper limit of the current warning threshold. The experiment determines whether the current value is a fault current, that is, whether the motor is faulty when the current middle value is used as the current value. If the current middle value is a fault current, the middle value between the current value and the rated current is continued to be taken, and the new middle value is verified for motor fault until the current value obtained by binary search is the upper limit of the current value for normal motor operation, and the current value of the critical value is used as the current warning threshold of the motor.

[0085] Step 3. For motor currents that do not exceed the threshold, further compare the lateral currents of the same type and multiple motors based on the KNN clustering algorithm, and further proceed to step 4 for judgment for motor currents with relatively small data differences.

[0086] The motor current with relatively large data difference is compared with the rated current value.

[0087] Determine whether the motor current is within the preset safety range. If the motor current is within the preset safety range, go to step 4 for further determination. If the motor current is not within the preset safety range, directly determine that the motor is abnormal.

[0088] Specifically, the current mean of multiple motors of the same type is first calculated, and then the distance between the current value of each motor of the same type and the obtained current mean of the type is further calculated:

[0089]

[0090] Wherein, n is the number of times the current data is obtained, and the current data includes the time and current value of the current comparison; x n Indicates the time of obtaining the current comparison for the nth time, represents the average time of obtaining current comparison n times, and 0.2 is the weight coefficient of the time of obtaining current comparison, which is used to reduce the error of overall distance judgment caused by the time difference of uploading from different smart gateways; n Indicates the current value obtained for the nth time, represents the average of the current values ​​obtained n times, and 0.8 is the weight coefficient of the current value.

[0091] Sort the calculated distance values ​​in ascending order.

[0092] The K distance values ​​that are closest to the front are selected as the motor setting K values ​​that participate in the lateral current comparison.

[0093] The first K motor currents obtained by arranging in ascending order are further compared with the longitudinal historical current data using step 4; the remaining motor currents after excluding the first K motor currents after arranging in ascending order are judged as follows:

[0094] Calculate whether the difference between the current value of the remaining motors and the rated current value is within a preset safety range;

[0095] For motor currents whose difference is within the safe range, the process proceeds to step 4 to continue the longitudinal historical current data comparison. For motor currents whose difference is outside the safe range, the process directly determines that the motor is abnormal.

[0096] It should be noted that motor failure corresponds to a major mechanical fault or line fault in the motor, which directly affects the use of the wind turbine system. Motor abnormality corresponds to a minor fault or line problem in the motor, which has little impact on current use, but it is necessary to pay attention to the development trend of the motor abnormality and replace or repair the motor in time to avoid affecting subsequent use.

[0097] The process of horizontal comparative cluster analysis is as follows Fig.10 shown.

[0098] The motor current data of the same model and type of motors in the same area of ​​the in-service wind turbine generator sets are compared. When all motors operate normally, the current operating curves of their motors tend to a certain mean value.

[0099] The KNN clustering algorithm is used to calculate the distance between the motor current value and the current mean, and the motor current data participating in the comparison are arranged in ascending order according to the distance difference, and then the K value is set.

[0100] For example, if there are 10 groups of motors involved in the comparison, the K value can be set to 7, then the current values ​​of the first 7 motors closest to the mean current (that is, the first 7 motors arranged in ascending order of distance value) are all regarded as normal motor currents.

[0101] The currents of the remaining three motors will be compared with the rated currents of the motors for difference analysis.

[0102] If the difference is within the preset safety range, it will continue to be used for subsequent longitudinal historical current data comparison. If the difference exceeds the safety range, it will be judged as a motor failure and an alarm message will be triggered to indicate the motor abnormality.

[0103] The key to the accuracy of KNN clustering algorithm in motor fault analysis and judgment lies in the setting of K value.

[0104] When the value of K is too small, once there is noise, it will have a greater impact on the motor fault judgment. For example, when the K value is 1, once the nearest point is noise, there will be a large deviation. The reduction of the K value means that the overall model becomes complex and prone to overfitting. When the value of K is too large, it is equivalent to using training examples in a larger neighborhood, and the approximate error of learning will increase. At this time, examples far away from the input target point will also play a role in the motor fault analysis and judgment, causing errors in the motor fault judgment. Therefore, the setting of the K value requires continuous data feeding and optimization.

[0105] The specific process of setting the K value is as follows Fig.11 As shown, the specific process is as follows:

[0106] First, a K value is temporarily determined based on the N motor currents participating in the same type and multi-motor lateral current comparison; the first K motor currents are analyzed to determine whether there are abnormal current values ​​in the first K motor currents. If there are abnormal current values, the K value is reduced until there are no abnormal current values ​​in the first K motor currents; if the first K motor currents are all normal current values, the remaining NK motor currents are analyzed again. If the normal current values ​​account for no less than 30% of the remaining NK motor currents, the K value is increased until the normal current values ​​account for less than 30% of the remaining NK current values.

[0107] Through the horizontal comparison and cluster analysis of multiple groups of motors, it is possible to analyze the circuit problems and potential motor failures of the motors, and also to analyze and determine the potential failures of motors that seem to be operating normally, such as loose wiring of the motors and aging of the motor circuits. The specific process is as follows: Fig.13 As shown, the motor roughly judged by the threshold indicates that the current motor is still operating normally; further, the line problems and potential motor failures of the motor are analyzed through horizontal comparison and cluster analysis of multiple groups of motors. If, in the process of horizontal comparison of motor current values, the current value of a certain motor is discrete from the current values ​​of other motors (specifically, the distance value between the current value of each motor of the same type and the current mean of the type obtained is calculated and sorted in ascending order according to each distance value, and the remaining motor currents after removing the first K motor currents) indicate that the motor current may be abnormal, and further analyze the three-phase currents of the motor A, B, and C. If the three-phase currents are all higher than the normal values ​​of other motor currents, it indicates that there is a minor mechanical fault problem in the internal mechanical structure of the motor; if the current value of one or two phases of the motor current is abnormal, it indicates that there is a loose wire or aging problem in the line of the motor.

[0108] Step 4. Compare the motor current entering the longitudinal historical current data comparison with its own motor current historical normal value. The average of its own motor current historical normal value is the average of the historical current data of the motor under normal operation.

[0109] The historical normal value of the motor current is the historical current data under normal operation of the motor. The process of comparing the longitudinal historical current data is as follows: Fig.12 If the difference between the motor current value and the average of the historical normal values ​​of the motor current is within the preset range, the motor is considered to be operating normally; if the difference between the motor current value and the average of the historical normal values ​​of the motor current is outside the preset range, the motor is considered to be abnormal; by further performing trend analysis on the difference between the motor current value and each historical normal value of the motor current, the specific situation of the motor abnormality can be judged based on whether the difference development trend changes steadily over time.

[0110] For example, if the trend is stable, it may be caused by dust on the motor or aging of the circuit. If the trend is rising or falling, it means that the problem circuit or internal mechanical problems are becoming more and more serious. The current longitudinal comparison analysis compares the current of the motor operation with the historical normal value to eliminate the misjudgment of motor failure caused by problems with the motor's own properties.

[0111] Specific as Fig.14 As shown, the mean value of the historical current data under normal operation of the motor is set as the mean value of the historical normal value of the motor current itself, and the current current value of the motor is compared with the mean value of the historical normal value of the motor current itself (the mean value is the mean value of the historical current data under normal operation of the motor, which is a constant after determination).

[0112] First, determine the average historical normal value of the motor current of the motor itself; then confirm the upper and lower fluctuation range of the motor current. Assuming that the normal upper and lower fluctuation range of the current is 0.5A, in order to prevent interference, the threshold of the fluctuation range is set to 0.8A. It is judged that the motor is operating normally if the fluctuation above and below the average historical normal value of the motor current does not exceed 0.8A. If the difference between the motor current value and the average historical normal value of the motor current is greater than 0.8A, it is judged as a motor abnormality.

[0113] For motors with abnormal motor current, current difference analysis is performed. When trend analysis is performed on abnormal motors, the difference of the normal historical motor current data of the motor will be saved. When performing difference analysis, the difference historical data before the abnormal situation occurs will also be retrieved for analysis. The motor current value is subtracted from the historical normal value of the motor current. If the difference data increases gradually over time, it means that the problem circuit or internal mechanical problem is becoming more and more serious. If the difference has been stable, it may be caused by dust or aging of the line, and the wind turbine needs to be cleaned and maintained in time.

[0114] The method of the present invention is based on real-time remote acquisition of motor current data by a current acquisition system, and the real-time analysis and processing of motor current data is realized through a cloud server. The current acquisition system is easy to install and wire on site, and has a high degree of feasibility. The present invention constructs a diversified motor fault warning method, which can judge motor operation failures and abnormal conditions based on current data analysis, compare the motor current data after filtering out the motor's starting current, and then compare the horizontal current of the same type and multiple motors based on the KNN clustering algorithm, as well as the longitudinal historical current data of a single motor, to grade and judge whether the motor operation status is normal.

[0115] This method is designed to use three levels of judgment: threshold comparison, horizontal current comparison of multiple motors of the same type, and vertical current comparison of single motor. First, using the threshold rough judgment can directly determine the motor with large mechanical failure or line problem. This motor has a serious impact on the system and must be replaced in time; the motors entering the horizontal comparison are all within the threshold, indicating that the motor can currently operate normally and has little impact on the current system operation. The horizontal comparison of multiple motors of the same type and the same area can be more detailed. Analysis and discovery of potential problems and failures in the operation of a motor, this type of motor can be replaced in time but needs to be continuously paid attention to; for the final vertical comparison, in addition to being able to see the healthy development and changes of the motor itself, it can also filter out false alarms caused by the motor's own characteristics in the horizontal comparison. If the current data of a motor itself is different from others, through the vertical comparison of the motor itself, it is found that it has been in this state, but the motor state is normal, indicating that the motor is a special case and should be filtered out during the horizontal comparison to ensure the reliability and stability of the three-level warning.

[0116] For example, in extreme cases, the current data of a motor in normal operation is higher than that of other motors, and the difference between its current value and the rated current is greater than the set safety range. In this case, the motor will always be judged as having abnormal current during horizontal comparison. Therefore, it is necessary to make special markings on it so that the motor no longer participates in the horizontal comparison of the motors selected in the area, and directly participates in the final vertical comparison link, thereby reducing the data interference caused by special data to the horizontal comparison analysis and improving the accuracy of the horizontal comparison analysis.

[0117] Example 2

[0118] This embodiment 2 describes an online detection system for wind turbine motor faults based on current sensing, which is based on the same inventive concept as the online detection method for wind turbine motor faults based on current sensing in embodiment 1.

[0119] Specifically, the wind turbine motor fault online detection system based on current sensing includes the following modules:

[0120] A data cleaning module is used to clean the current data of the corresponding motor uploaded by each current transformer in the current acquisition system received by the cloud server to filter out the starting current of the motor;

[0121] The threshold judgment module performs a rough threshold judgment on the motor current data after data cleaning;

[0122] When the motor current value collected by the current transformer exceeds the preset threshold, it is determined that the motor is faulty; when the motor current value collected by the current transformer does not exceed the preset threshold, it switches to the lateral current comparison module of multiple motors of the same type;

[0123] The same type of multi-motor transverse current comparison module is used to further compare the same type and multi-motor transverse currents based on the KNN clustering algorithm for motor currents that do not exceed the threshold;

[0124] The motor current with relatively small data difference is transferred to the longitudinal historical current data comparison module for judgment; the motor current with relatively large data difference is compared with the rated current value to determine whether it is within the preset safety range;

[0125] If the motor current is within the preset safety range, it will be transferred to the longitudinal historical current data comparison module for further judgment. If the motor current is not within the preset safety range, it will be directly judged as a motor abnormality;

[0126] and a longitudinal historical current data comparison module, which is used to compare the motor current value with its own historical normal value of the motor current; the historical normal value of its own motor current is the historical current data under normal operation of the motor;

[0127] If the difference between the motor current value and the average of the motor current's historical normal values ​​is within the preset range, the motor is deemed to be operating normally; otherwise, the difference between the motor current value and its historical normal values ​​is further determined.

[0128] The specific situation of the motor abnormality can be further distinguished based on whether the difference development trend changes smoothly over time.

[0129] It should be noted that, in the online detection system for motor faults of wind turbines based on current sensing, the implementation process of the functions and effects of each functional module is specifically described in the implementation process of the corresponding steps in the method in Example 1, which will not be repeated here.

[0130] Example 3

[0131] This embodiment 3 describes a computer device, which includes a memory and one or more processors.

[0132] An executable code is stored in the memory. When the processor executes the executable code, it is used to implement the steps of the online detection method for wind turbine motor fault based on current sensing in the above-mentioned embodiment 1.

[0133] In this embodiment, the computer device is any device or apparatus with data processing capability, which will not be described in detail here.

[0134] Example 4

[0135] This embodiment 4 describes a computer-readable storage medium on which a program is stored. When the program is executed by a processor, the program is used for the steps of an online detection method for wind turbine motor faults based on current sensing.

[0136] The computer-readable storage medium can be an internal storage unit of any device or apparatus with data processing capabilities, such as a hard disk or memory, or an external storage device of any device with data processing capabilities, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), an SD card, a flash card, etc. equipped on the device.

[0137] Of course, the above description is only a preferred embodiment of the present invention, and the present invention is not limited to the above embodiments. It should be noted that all equivalent substitutions and obvious deformation forms made by any technician familiar with the field under the guidance of this specification fall within the essential scope of this specification and should be protected by the present invention.

Claims

1. A method for online detection of wind turbine motor faults based on current sensing, implemented based on a pre-built current acquisition system; characterized in that: The current acquisition system includes a cloud server and multiple current transformers; Each current transformer collects the current data of one phase of the three-phase current of any motor in the wind turbine generator, and the collected current data is uploaded to the cloud server via the intelligent gateway; The wind turbine generator motor fault online detection method based on current sensing comprises the following steps: Step 1. The cloud server receives the current data of the corresponding motor uploaded by each current transformer in the current acquisition system, and performs data cleaning on the current data to filter out the starting current of the corresponding motor; Step 2. Perform a rough threshold judgment on the motor current data after the data cleaning in step 1; When the motor current value collected by the current transformer exceeds the preset threshold, it is directly determined that the motor is faulty; when the motor current value collected by the current transformer does not exceed the preset threshold, go to step 3; Step 3. Compare the lateral currents of the same type and multiple motors based on KNN clustering for the motor currents that do not exceed the threshold; For motor currents with relatively small data differences, go to step 4 to compare longitudinal historical current data; The motor current with relatively large data difference is compared with the motor rated current to determine whether the motor current is within the preset safety range; If the motor current value is within the preset safety range, the process proceeds to step 4 to compare the motor current with the historical current data. If it is not within the preset safety range, the motor is judged to be abnormal. Step 4. Further compare the motor current value with its own motor current historical normal value, wherein the average of its own motor current historical normal value is the average of the historical current data of the motor under normal operation; If the difference between the motor current value and the average of the motor current's historical normal values ​​is within the preset range, the motor is deemed to be operating normally; otherwise, the difference between the motor current value and its historical normal values ​​is further determined. The specific situation of the motor abnormality can be further distinguished based on whether the difference development trend changes smoothly over time.

2. The method for online detection of wind turbine motor faults based on current sensing according to claim 1 is characterized in that: The current acquisition system includes a bottom sensing layer device and a cloud server; The bottom sensing layer equipment includes a current transmitter and multiple current transformers, each of which is used to collect one-phase current data of a motor in a wind turbine; The motor current data is converted from analog data to digital data by a current transmitter, and the collected motor current data is uploaded to the cloud server via wireless transmission.

3. The method for online detection of wind turbine motor faults based on current sensing according to claim 1 is characterized in that: In the step 1, the motor current data collected by the current collection system is judged, and when it is identified that the motor current changes from 0A to non-0A, the data within 2s of the current motor current is defined as the starting current; The starting current data is filtered, and the current motor current data after 2s is used for the rough threshold judgment of step 2.

4. The method for online detection of wind turbine motor faults based on current sensing according to claim 1 is characterized in that: In step 2, the method for determining the preset threshold is: Combined with the rated current of the motor itself and the historical data of the motor current collected by the intelligent gateway, the current value during the motor failure period, i.e., the motor fault current, is taken out, and the rated current of the motor and the critical value of the motor fault current are obtained by binary search, and the critical value is set as the current warning threshold, i.e., the preset threshold in the threshold rough judgment.

5. The method for online detection of wind turbine motor faults based on current sensing according to claim 1, characterized in that: The step 3 is specifically as follows: First, the current mean of multiple motors of the same type is calculated, and then the distance between the current value of each motor of the same type and the obtained current mean of the type is calculated, and the calculated distance values ​​are sorted in ascending order; Take the K values ​​that are the closest to the front among all the distance values ​​as the motor setting K values ​​that participate in the horizontal current comparison; The first K motor currents obtained by arranging in ascending order are further compared with the longitudinal historical current data using step 4; the remaining motor currents after excluding the first K motor currents after arranging in ascending order are judged as follows: Calculate whether the difference between the current value of the remaining motors and the rated current value is within a preset safety range; For motor currents whose difference is within the safe range, the process proceeds to step 4 to continue the longitudinal historical current data comparison. For motor currents whose difference is outside the safe range, the process directly determines that the motor is abnormal.

6. The method for online detection of wind turbine motor faults based on current sensing according to claim 5 is characterized in that: In step 3, the formula for calculating the distance between the current value of each motor of the same type and the current mean value of the type obtained is as follows: Wherein, n is the number of times the current data is obtained, and the current data includes the time and current value of the current comparison; x n Indicates the time of obtaining the current comparison for the nth time, represents the average time of obtaining current comparison n times, and 0.2 is the weight coefficient of the time of obtaining current comparison, which is used to reduce the error of overall distance judgment caused by the time difference of uploading from different smart gateways; n Indicates the current value obtained for the nth time, represents the average of the current values ​​obtained n times, and 0.8 is the weight coefficient of the current value.

7. The method for online detection of wind turbine motor faults based on current sensing according to claim 5 is characterized in that: In step 3, the process of adjusting the K value is as follows: First, a K value is tentatively determined based on the currents of N motors participating in the lateral current comparison of multiple motors of the same type; Analyze the first K motor currents to determine whether there are abnormal current values ​​in the first K motor currents. If there are abnormal current values, reduce the K value until there are no abnormal current values ​​in the first K motor currents. If the first K motor currents are all normal current values, analyze the remaining NK motor currents. If the normal current values ​​account for no less than 30% of the remaining NK motor currents, increase the K value until the normal current values ​​account for less than 30% of the remaining NK current values.

8. The online detection system for wind turbine motor fault based on current sensing is characterized by: include: A data cleaning module is used to clean the current data of the corresponding motor uploaded by each current transformer in the current acquisition system received by the cloud server to filter out the starting current of the motor; A threshold judgment module is used to make a rough threshold judgment on the motor current data after data cleaning; When the motor current value collected by the current transformer exceeds the preset threshold, it is determined that the motor is faulty; when the motor current value collected by the current transformer does not exceed the preset threshold, it switches to the lateral current comparison module of multiple motors of the same type; The same type of multi-motor transverse current comparison module is used to further compare the same type and multi-motor transverse currents based on the KNN clustering algorithm for motor currents that do not exceed the threshold; The motor current with relatively small data difference is transferred to the longitudinal historical current data comparison module for judgment; the motor current with relatively large data difference is compared with the rated current value to determine whether it is within the preset safety range; If the motor current is within the preset safety range, it will be transferred to the longitudinal historical current data comparison module for further judgment. If the motor current is not within the preset safety range, it will be directly judged as a motor abnormality; and a longitudinal historical current data comparison module, which is used to compare the motor current value with its own motor current historical normal value; the average of its own motor current historical normal value is the average of the historical current data of the motor under normal operation; If the difference between the motor current value and the average of the motor current's historical normal values ​​is within the preset range, the motor is deemed to be operating normally; otherwise, the difference between the motor current value and its historical normal values ​​is further determined. The specific situation of the motor abnormality can be further distinguished based on whether the difference development trend changes smoothly over time.

9. A computer device comprising a memory and one or more processors, wherein the memory stores executable code, characterized in that: When the processor executes the executable code, the steps of the method for online detection of wind turbine motor faults based on current sensing as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a program stored thereon, characterized in that: When the program is executed by the processor, the steps of the online detection method for wind turbine motor fault based on current sensing as described in any one of claims 1 to 7 are implemented.

Images