Method for anomaly detection of a variable pitch system, storage medium and computing device

Abstract

An abnormality detection method of a variable pitch system, a storage medium and a computing device are disclosed. The abnormality detection method comprises: obtaining operation data of each variable pitch motor of the variable pitch system during the variable pitch system performing a variable pitch operation; performing cumulative summation on the operation data of each variable pitch motor at each sampling time within a predetermined time period to obtain a data sum value at each sampling time; performing data consistency judgment on the data sum value at each sampling time; in response to the data sum value at each sampling time having consistency, determining that the variable pitch system is operating normally; and in response to the data sum value at each sampling time not having consistency, determining that the variable pitch system has an abnormality. In this way, the problem that the SCADA operation data sampling period is long, there is no continuous data, and effective monitoring cannot be achieved can be solved, and the problem that existing spectrum analysis, jump detection, amplitude detection and the like cannot correctly analyze the operating state of the wind turbine generator set can be solved.

Description

Technical Field

[0001] This disclosure generally relates to the field of wind power generation technology, and more specifically, to a method for detecting anomalies in a pitch system, a storage medium, and a computing device. Background Technology

[0002] With the advancement of technology, modern enterprises' electromechanical equipment is rapidly and comprehensively developing towards large-scale, continuous, and automated production, meeting the objective requirements of reducing production costs, increasing productivity, saving energy, reducing labor, minimizing waste, and improving product quality. The stable operation of various machines and equipment directly affects the economic benefits of enterprises; some critical equipment even plays a decisive role in the fate of the enterprise. Therefore, how to avoid equipment failures, especially catastrophic failures, has always been a highly valued issue. For a long time, due to the unpredictability of accidents, people have had to adopt the following two countermeasures.

[0003] One approach is to repair the equipment only after it has completely failed. This method results in significant economic losses because repairing equipment after it has completely failed often requires expensive repair costs and can lead to "under-repair," causing catastrophic damage. At best, it requires replacing the equipment; at worst, it can result in personal injury or death.

[0004] Secondly, regular equipment maintenance is essential. This method has a degree of planning and prevention, but its drawback is its high degree of randomness. If equipment is not malfunctioning, downtime can cause significant economic losses. This method often leads to "over-maintenance," resulting in increased costs and even waste of components.

[0005] Therefore, a reasonable maintenance approach should be predictive, meaning that potential problems can be detected and predicted early in the course of equipment failure, allowing for timely and appropriate measures to be taken. This approach enables timely and targeted maintenance, ensuring that the equipment is in good operating condition and making full use of the lifespan of parts, effectively avoiding both "under-maintenance" and "over-maintenance."

[0006] In the wind power industry, effectively providing early warnings of anomalies in key components of wind turbine generators is of great practical significance for the safe operation of wind turbine generators and the reduction of maintenance costs. Generally speaking, data acquisition is a prerequisite and foundation for anomaly (fault) early warning. However, existing data files and analysis methods in the wind power field have some shortcomings.

[0007] First, although the sampling period of the fault file is 20ms, which can obtain the raw data and data change trend to the greatest extent, the disadvantage is that the B file that records the operating environment status information can only be generated when the wind turbine generator fails and stops, and the duration is generally only 1 minute. Therefore, it is not convenient to monitor the wind turbine generator in real time and provide fault early warning, nor can it achieve continuous monitoring of the wind turbine generator 24 hours a day.

[0008] Secondly, while continuous data acquisition can obtain long-term 20ms sampling data, this type of continuous data acquisition is generally more suitable for data collection and fault analysis of a single wind turbine generator. This is because data acquisition requires hardware devices such as PCs or industrial control computers to collect data from all wind turbine generators in a wind farm. This often necessitates configuring multiple industrial control computers, resulting in high costs, long processing times, and a high risk of data acquisition interruptions.

[0009] Third, although instantaneous data files can be exported from the SCADA system and contain continuous data, the sampling period of the SCADA system is relatively long, typically recording a data value every 5 to 7 seconds. Due to this long sampling period, existing analysis methods (such as spectrum analysis, fluctuation detection, jump detection, peak detection, and margin detection) cannot correctly interpret the operating status of wind turbine generators.

[0010] Fourth, the built-in algorithm developed within the PLC controller can output the detection results to SCADA. However, this built-in method has several drawbacks: first, it requires upgrading the controller's program, and the data before the upgrade cannot be analyzed; second, adjusting the detection parameters is inconvenient, and the data before the parameter adjustment cannot be analyzed; and third, the data only has fixed output results, making the database inflexible.

[0011] Figure 1 This is a waveform diagram showing the data fluctuations recorded in the B file and the instantaneous data file. For example... Figure 1 As shown, the left side of the graph displays the waveform of data fluctuations recorded in file B, while the right side displays the waveform of data fluctuations recorded in the instantaneous data file. From Figure 1 It can be seen that although the data recorded in the B file with a sampling period of 20ms fluctuates greatly, only one data value shown in the red box can be collected in the instantaneous file with a sampling period of 7s. That is, it is difficult to restore the trend of data change in the instantaneous data file. Summary of the Invention

[0012] Therefore, embodiments of this disclosure provide an anomaly detection method, storage medium, and computing device for a pitch system, which can perform anomaly diagnosis and early warning based on SCADA operating data. On the one hand, it solves the problems of long sampling period, lack of continuous data, and inability to achieve effective monitoring of SCADA operating data. On the other hand, it solves the problem that existing spectrum analysis, jump detection, amplitude detection, etc., cannot correctly interpret the operating status of wind turbine generators.

[0013] In one general aspect, an anomaly detection method for a pitch system is provided. The anomaly detection method includes: acquiring the operating data of each pitch motor of the pitch system during pitch operation; summing the operating data of each pitch motor at each sampling time within a predetermined time period to obtain the data and value at each sampling time; determining the data consistency of the data and value at each sampling time; determining that the pitch system is operating normally in response to the consistency of the data and value at each sampling time; and determining that an anomaly exists in the pitch system in response to the inconsistency of the data and value at each sampling time.

[0014] Optionally, in response to the inconsistency of data and values ​​at each sampling time, the step of determining that there is an anomaly in the pitch system includes: determining whether the anomaly in the pitch system is caused by an anomaly in the pitch bearing or by an anomaly in the blades of the wind turbine generator based on the data and values ​​at each sampling time.

[0015] Optionally, the anomaly detection method further includes: acquiring the rotor speed of the wind turbine generator during the pitch control operation of the pitch system; calculating the rotor rotation period based on the rotor speed; and, in response to the discrepancy between the rotor rotation period and the sampling period in time, performing a step of summing up the operating data of each pitch motor at each sampling moment within a predetermined time period.

[0016] Optionally, the step of determining the data consistency of data and values ​​at each sampling time includes: determining whether the data and values ​​at each sampling time are all less than a first preset threshold; and determining that the data and values ​​at each sampling time are consistent in response to the fact that the data and values ​​at each sampling time are all less than the first preset threshold.

[0017] Optionally, the step of determining the data consistency of the data and values ​​at each sampling time further includes: in response to the data and values ​​at at least one sampling time being greater than or equal to the first preset threshold, calculating the difference between the data and values ​​at each sampling time and the data and values ​​at the previous sampling time to obtain the difference between the data and values ​​at each sampling time, wherein each sampling time includes all sampling times other than the first sampling time within a predetermined time period; and determining the data consistency of the data and values ​​at each sampling time based on the difference between the data and values ​​at each sampling time.

[0018] Optionally, the step of determining the data consistency of the data and values ​​at each sampling time includes: calculating the difference between the data and values ​​at each sampling time and the data and values ​​at the previous sampling time to obtain the difference between the data and values ​​at each sampling time, wherein each sampling time includes all sampling times except the first sampling time within a predetermined time period; and determining the data consistency of the data and values ​​at each sampling time based on the difference between the data and values ​​at each sampling time.

[0019] Optionally, the step of determining the data consistency of data and values ​​at each sampling time based on the difference between the data and values ​​at each sampling time includes: for each sampling time, determining whether the absolute value of the difference between the data and values ​​at the corresponding sampling time is less than a second preset threshold, and in response to the absolute value of the difference between the data and values ​​at the corresponding sampling time being less than the second preset threshold, accumulating a first count value; and in response to the first count value being greater than the first count threshold, determining that the data and values ​​at each sampling time are consistent.

[0020] Optionally, the step of determining data consistency based on the difference between data and values ​​at each sampling time includes: for each sampling time, determining whether the absolute value of the difference between data and values ​​at the corresponding sampling time is greater than or equal to a second preset threshold, and in response to the absolute value of the difference between data and values ​​at the corresponding sampling time being greater than or equal to the second preset threshold, accumulating a second count value; and in response to the second count value being greater than the second count threshold, determining that the data and values ​​at each sampling time are not consistent.

[0021] Optionally, the step of determining that there is an anomaly in the pitch system in response to the inconsistency of data and values ​​at each sampling time includes: determining that the anomaly in the pitch system is caused by an anomaly in the pitch bearing in response to the second count value being greater than the second count threshold.

[0022] Optionally, the step of determining data consistency for the data and values ​​at each sampling time based on the difference between the data and values ​​at each sampling time includes: for each sampling time, determining whether the sign of the difference between the data and values ​​at the corresponding sampling time and the second preset threshold is different from the sign of the difference between the data and values ​​at the previous sampling time and the second preset threshold; and in response to the difference between the sign of the difference between the data and values ​​at the corresponding sampling time and the second preset threshold being different from the sign of the difference between the data and values ​​at the previous sampling time and the second preset threshold, accumulating a third count value; and in response to the third count value being greater than the third count threshold, determining that the data and values ​​at each sampling time are not consistent.

[0023] Optionally, the step of determining that the pitch system has an anomaly in response to the inconsistency of data and values ​​at each sampling time includes: determining whether the sampling times for accumulating the third count value are regularly distributed in time in response to the inconsistency of data and values ​​at each sampling time; determining that the anomaly of the pitch system is caused by an anomaly of the wind turbine blades in response to the regularly distributed sampling times for accumulating the third count value in time; and determining that the anomaly of the pitch system is caused by an anomaly of the pitch bearing in response to the irregular distribution of the sampling times for accumulating the third count value in time.

[0024] Optionally, the operating data of each pitch motor is SCADA operating data, and includes at least one of the torque value, current value, given speed value and actual speed value of each pitch motor.

[0025] In another general aspect, a storage medium storing a computer program is provided, which, when executed by a processor, implements the anomaly detection method for a pitch system as described above.

[0026] In another general aspect, a computing device is provided, the computing device comprising: a processor; and a memory storing a computer program that, when executed by the processor, implements the anomaly detection method for the pitch system as described above.

[0027] The anomaly detection method, storage medium, and computing device of the pitch system according to the embodiments of this disclosure can realize anomaly diagnosis and early warning based on SCADA operation data, effectively solve the problem of excessively long SCADA operation data sampling period, realize the correct parsing of wind turbine operation data that cannot be achieved by existing spectrum analysis, jump detection, amplitude detection, etc., and improve the application value of operation data.

[0028] On the other hand, the anomaly detection method, storage medium, and computing device of the pitch system according to the embodiments of this disclosure do not require upgrading or optimizing the PLC program, nor do they require repeated adjustments to the detection parameters. They are particularly suitable for SCADA transient data files generated from historical dates, which greatly improves the effectiveness and reliability of anomaly detection.

[0029] Furthermore, the anomaly detection method, storage medium, and computing device for the pitch system according to embodiments of this disclosure are minimally affected by torque value fluctuations because they do not involve the detection of spectrum or amplitude jumps, effectively reducing the probability of false detections. Simultaneously, they solve the problem of a small number of B files (generated only during shutdown), enabling 24-hour monitoring of the pitch system's operating status; and they do not require modification of SCADA programs or communication protocols between the wind turbine generator and central monitoring, thus facilitating implementation and reducing development workload. Attached Figure Description

[0030] The above and other objects and features of the embodiments of this disclosure will become clearer from the following description taken in conjunction with the accompanying drawings illustrating the embodiments, wherein:

[0031] Figure 1 It is a waveform diagram showing the data fluctuations recorded in the B file and the instantaneous data file;

[0032] Figure 2 It is a diagram showing the waveform and vector decomposition of a standard three-phase AC signal;

[0033] Figure 3 A force analysis diagram of the blade is shown in the circular region where the blade rotates.

[0034] Figure 4 It is a diagram showing the torque curves of the three pitch motors when the wind turbine is pitch-tuning;

[0035] Figure 5 It is a diagram showing the torque curves of the three pitch motors when the wind turbine is pitch-tuning;

[0036] Figure 6 It is a diagram showing the torque curves of the three pitch motors when the wind turbine is running and shut down;

[0037] Figure 7 This is a diagram showing the torque curve of the pitch system when the blades malfunction;

[0038] Figure 8 It is a graph showing the changes in three-phase data;

[0039] Figure 9 This is a flowchart illustrating an anomaly detection method for a pitch system according to an embodiment of the present disclosure;

[0040] Figure 10 It is a graph showing the current values ​​of the three pitch motors when the pitch bearing is malfunctioning;

[0041] Figure 11 It is a waveform diagram showing the data and values ​​at each sampling time;

[0042] Figure 12 This is a block diagram illustrating a computing device according to an embodiment of the present disclosure. Detailed Implementation

[0043] The following detailed embodiments are provided to aid the reader in gaining a comprehensive understanding of the methods, apparatus, and / or systems described herein. However, various changes, modifications, and equivalents of the methods, apparatus, and / or systems described herein will become apparent upon understanding this disclosure. For example, the order of operations described herein is merely illustrative and is not limited to those orders set forth herein, but may be changed as will become clear upon understanding this disclosure, except for operations that must occur in a specific order. Furthermore, for clarity and conciseness, descriptions of features known in the art may be omitted.

[0044] The implementation principle of the anomaly detection method for a pitch system according to an embodiment of the present disclosure is described below.

[0045] Figure 2 It is a diagram showing the waveform and vector decomposition of a standard three-phase AC signal.

[0046] Reference Figure 2 When the three-phase load is balanced, that is, when the resistance of the three-phase load is the same, the three-phase current will be the same. Let the resistance be R, then the instantaneous voltages of the three phases are usinα, usin(α+2 / 3π), and usin(α+4 / 3π). u is the maximum voltage, α is any electrical angle, which can be understood as any time, π is the angle in radians, 2 / 3π is 120 degrees, and sin is the sine function.

[0047] The three-phase currents are: usinα / R, usin(α+2 / 3π) / R, and usin(α+4 / 3π) / R.

[0048] The sum of the currents is usinα / R + usin(α + 2 / 3π) / R + usin(α + 4 / 3π) / R, and the sum of the currents can be further derived as follows.

[0049]

[0050] The final result is 0 amperes, meaning the sum of the three-phase currents is 0.

[0051] Table 1 shows the data (e.g., current, etc.) of a standard three-phase electrical system. As shown in the "AB Difference" column of Table 1, the maximum absolute value of the difference is 1.73. Therefore, accurate condition detection cannot be achieved solely by judging the difference between two-phase data. However, it can be seen from Table 1 that the sum of the three-phase data is 0 at any given time. Therefore, this disclosure utilizes this principle to study the operating mechanism and characteristics of wind turbine generator sets and develops an anomaly diagnosis and early warning method based on SCADA operating data, effectively solving the problems of long sampling time, lack of continuous data, and inability to achieve effective monitoring in existing technologies.

[0052] Table 1

[0053]

[0054]

[0055] The operating characteristics of wind power units will be analyzed in detail below.

[0056] Figure 3 The diagram shows the force analysis of the wind turbine blades. (Example:) Figure 3 As shown, mass 204 is the equivalent mass m when blade 107 is mounted on the hub and rotates; the first component mass 205 is the component mass of mass 204 in the direction perpendicular to blade 107; the second component mass 206 is the component mass of mass 204 in the direction parallel to the rotation axis 304. Mass 204, the first component mass 205, and the second component mass 206 are all located in the vertical plane formed by the three blades 107, and the angle formed by the directions of mass 204 and the second component mass 206 is α (degrees).

[0057] Figure 4 This diagram shows the force analysis of the blade in the circular region where the blade rotates. (See diagram for example.) Figure 4 As shown, the third mass component 302 is the mass component of the first mass component 205 in the direction tangent to the circle in which the blade 107 rotates, and the fourth mass component 303 is the mass component of the first mass component 205 in the direction perpendicular to the blade axis (i.e., Figure 4 The component mass in the direction of the rotation axis 305). From the... Figure 4 It can be seen that the first mass component 205 is perpendicular to the blade side surface, but not perpendicular to the rotation axis 304; the third mass component 302 is perpendicular to the rotation axis 304.

[0058] Furthermore, by Figure 3 It can be seen that the size of the first component mass 205 is equal to the mass 204 multiplied by sina. Assuming the vertically upward position is 0 degrees, then the azimuth angle of the blade 107 at the 0-degree position, measured by the azimuth sensor 112 facing the hub 101, is set as a1 (degrees). Figure 3 It can be seen that a1 is 0; in the state of Figure 3 The azimuth angle of blade 107 on the right half is set as a2 (degrees), where a2 is a1 + 120 (degrees); [This is in the context of...] Figure 3 The azimuth angle of blade 107 on the left half is set to a3 (degrees), where a3 is a1 + 240 (degrees). Figure 3 In each of the blades 107 shown, since a1 = 0, a2 = 120 and a3 = 240; then the angle between the direction of mass 204 and the direction of the second mass 206 can be calculated as a = 180 - 120 = 60 degrees.

[0059] On the other hand, by Figure 4It can be seen that the size of the third mass 302 is equal to the first mass 205 multiplied by sinb, where b is the blade angle (i.e., the pitch angle) measured by the encoder, which is the angle between the blade side and the vertical plane 301 formed by the three blades 107 (e.g., ...). Figure 4 As shown), Figure 4 The vertical plane shown is Figure 3 They are on the same plane.

[0060] Assuming the initial position of blade 107 is 90 degrees, when blade 107 is located in the right half of the vertical plane 301 formed by the three blades 107 (i.e., Figure 3 When the blade 107 is located in the right half of the plane of the vertical plane 301 and the pitch direction is pitching towards 0 degrees, the additional moment of inertia I2 of the blade 107 acts as a drag on the pitch; when the blade 107 is located in the right half of the vertical plane 301 and the pitch direction is pitching towards 180 degrees, the additional moment of inertia I2 of the blade 107 acts as an assist on the pitch.

[0061] When blade 107 is located in the left half-plane of vertical plane 301 (i.e. Figure 3 When the blade 107 is located in the left half of the plane of the vertical plane 301 and the pitch direction is pitching towards 0 degrees, the additional moment of inertia I2 of the blade 107 assists the pitching; when the blade 107 is located in the left half of the vertical plane 301 and the pitch direction is pitching towards 180 degrees, the additional moment of inertia I2 of the blade 107 acts as a drag on the pitching.

[0062] Therefore, the magnitude of the additional moment of inertia of blade 107 can be obtained as follows:

[0063] I2 = R 2 ×m×sina×sinb×p×(a-180) / |(a-180)| (1)

[0064] Where p represents the pitch direction of blade 107. When blade 107 pitches towards the 0-degree position, p = +1. When blade 107 pitches towards the 180-degree position, p = -1.

[0065] At the same time, the magnitude N of the torque caused by the gravity of blade 107 can be obtained as follows:

[0066] N = m×sina×sinb×p×(a-180) / |(a-180)| (2)

[0067] As shown in equations (1) and (2), when the nacelle of a wind turbine deviates from the wind direction by an angle α, the lift force on the three blades of the wind turbine is different. Furthermore, the blades have flanges and chords, resulting in uneven force distribution and increased load on the turbine. When the wind direction forms an acute angle with the blade rotation direction, the right half of the turbine experiences greater drag and less lift when facing the turbine. When the wind direction forms an obtuse angle with the blade rotation direction, the left half of the turbine experiences greater drag and less lift. Here, lift refers to the power of the wind driving the blades to rotate in the direction of rotor rotation; drag refers to the load force experienced by the blades during pitch adjustment.

[0068] Figure 5 This is a diagram showing the torque curves of the three pitch motors during wind turbine pitch control. Curves 501, 502, and 503 are the torque curves of the three pitch motors, respectively. From... Figure 5 As can be seen, under normal pitch control conditions, the torque of the three pitch motors exhibits sinusoidal fluctuations with a phase difference of 120 degrees. This indicates that the torque value of the pitch motor changes periodically with the rotation of the impeller.

[0069] Figure 6 This is a diagram showing the torque curves of the three pitch motors of the wind turbine generator set during operation and shutdown. Curves 601, 602, and 603 are the torque curves of the three pitch motors, respectively. Figure 6 As can be seen, under normal pitch control conditions, the torque of the three pitch motors exhibits sinusoidal fluctuations, and as the speed of the wind turbine rotor decreases, the period of its sinusoidal fluctuations gradually changes.

[0070] based on Figures 2 to 6 This further provides a basis for the feasibility of this disclosure. Specifically, since the speed of the pitch control system of a wind turbine generator has two directions, positive and negative, which correspond to the direction of rotation of the pitch motor, a positive speed indicates that the pitch control system is retracting (pitching towards 90 degrees to the blade) and a positive speed indicates that the pitch control system is opening (pitching towards 0 degrees to the blade). Therefore, for the wind turbine generator and the pitch drive, the torque and current values ​​also have positive and negative values.

[0071] Figure 7 This is a diagram showing the torque curve of the pitch system when the blades malfunction. From Figure 7 As can be seen, due to the significant resistance experienced by the blades and pitch motor corresponding to torque 703, its torque value is significantly higher than that of the other two shafts, and its fluctuation period is (822-355)*20ms = 9340ms = 9.34s, which is greater than the data sampling period of the SCADA system (5-7 seconds). Therefore, this phenomenon can be effectively identified in the SCADA operating data using the anomaly detection method of the pitch system according to the embodiments of this disclosure.

[0072] In addition, Figure 7 While it might seem that anomalies can be identified through the difference between the three torques, the torque values ​​of the three pitch motors are sinusoidal and have a 120-degree phase difference. Therefore, under normal circumstances, the difference between any two torque values ​​could be quite large (see the analysis in Table 1). Furthermore, Figure 8 It is a graph showing the changes in three-phase data. For example... Figure 8 As shown, if the data sampling point happens to be at time t1, t2 or t3, the difference between the two phases will be relatively large, which will greatly interfere with anomaly detection and may even lead to false detection.

[0073] Compared with the method of comparing the differences in the operating data of three pitch motors, the advantages and feasibility of the anomaly detection method of the pitch system according to the embodiments of this disclosure are also reflected in the following: since the loads of the pitch motors are different, the current values ​​and torque values ​​are different, and the loads are greatly related to wind speed, wind direction, gear oil temperature, etc., comparing the differences alone cannot effectively identify anomalies.

[0074] The following describes in detail an anomaly detection method for a pitch system according to embodiments of the present disclosure.

[0075] Figure 9 This is a flowchart illustrating an anomaly detection method for a pitch system according to an embodiment of the present disclosure. According to embodiments of the present disclosure, the anomaly detection method for the pitch system can be executed by a SCADA industrial control computer, data analysis platform, or other computing device located in the wind farm control center. However, the present disclosure is not limited thereto; the anomaly detection method for the pitch system can also be executed by a host computer located outside the wind farm, or even by the master controller of a single wind turbine generator, as long as the executing device can acquire SCADA operating data.

[0076] Reference Figure 9 In step S901, during the pitch operation of the wind turbine generator's pitch system, operating data of each pitch motor can be acquired. As described above, the operating data of each pitch motor can be SCADA operating data and includes at least one of the torque value, current value, given speed value, and actual speed value of each pitch motor. Thus, according to embodiments of this disclosure, various operating data of the pitch motors can be used for anomaly detection, thereby improving the usability of the anomaly detection method for the pitch system.

[0077] According to embodiments of this disclosure, the angle values ​​of each blade can also be obtained, and it can be determined whether the wind turbine generator set is in a pitch control state (including wind turbine generator set startup, power-limited pitch control, speed-limited pitch control, etc.) based on the angle values, that is, whether it is during pitch operation. For example, when the angle value is not 0, it can indicate that the wind turbine generator set is in a pitch control state. Optionally, if the pitch system has a 0-speed hold function (i.e., the torque of the pitch motor is not 0 when the pitch is not changed), it can be determined whether the wind turbine generator set is in a pitch control state by the torque value of the pitch motor. However, this disclosure is not limited to this, and various other existing methods can also be used to determine whether the generator set is in a pitch control state. When it is determined that the wind turbine generator set is in a pitch control state, the abnormal detection method of the pitch system according to embodiments of this disclosure can be executed.

[0078] In step S902, the operating data of each pitch motor at each sampling moment within a predetermined time period can be summed to obtain the data and value for each sampling moment. For example, at each sampling moment, the torque values ​​(or current values, given speed values, and actual speed values) of the three pitch motors can be summed to obtain the data and value for each sampling moment. The duration of the predetermined time period can be, for example, 5 minutes, but this disclosure is not limited thereto. The duration of the predetermined time period can be greater than or less than 5 minutes.

[0079] According to embodiments of this disclosure, the rotor speed of the wind turbine can also be acquired during the pitch control operation of the pitch system, and the rotor rotation period can be calculated based on the rotor speed. Subsequently, step S902 can only be executed if it is determined that the rotor rotation period and the sampling period are inconsistent in time. That is, if the rotor rotation period and the sampling period are approximately consistent in time (i.e., the start and end points of the period approximately coincide in time), the anomaly detection method for the pitch system ends, and anomaly detection is temporarily suspended. This is because if the rotor rotation period and the sampling period are approximately consistent in time, it means that at the two data sampling times, the three blades are at the same azimuth angle position as in the previous sampling, and the data from the two samplings are essentially the same, reducing the data's reference value. Therefore, by performing a consistency judgment between the rotor rotation period and the sampling period, the reliability of the operating data can be ensured, thereby reducing the probability of false detection.

[0080] In step S903, a data consistency judgment can be performed on the data and values ​​at each sampling time.

[0081] In step S904, the pitch system is determined to be operating normally if the data and values ​​at each sampling time are consistent. In step S905, an anomaly in the pitch system is determined if the data and values ​​at each sampling time are inconsistent. Further, in step S905, it can be determined whether the anomaly in the pitch system is caused by a pitch bearing malfunction or a blade malfunction of the wind turbine generator, based on the data and values ​​at each sampling time. This allows for more accurate detection of pitch bearing malfunctions (e.g., pitch bearing jamming) and blade malfunctions (e.g., blade breakage), thereby improving the usability of the pitch system anomaly detection method.

[0082] The following is a detailed description of the data consistency judgment.

[0083] According to an embodiment of this disclosure, in step S903, it can be determined whether the data and values ​​at each sampling time are all less than a first preset threshold. If the data and values ​​at each sampling time are all less than the first preset threshold, it is determined that the data and values ​​at each sampling time are consistent. Here, by determining whether the data and values ​​at each sampling time are all less than the first preset threshold, it can be determined whether the data and values ​​at each sampling time are close to 0. For example, the first preset threshold can be set to 0.5, but this disclosure is not limited thereto. Those skilled in the art can arbitrarily set the first preset threshold according to actual needs, as long as it is ensured that it can be determined whether the data and values ​​at each sampling time are close to 0. As shown in Table 1, at any time, the sum of the three-phase data is 0. If the data and values ​​are continuously close to 0, it indicates that the operating data of each pitch motor conforms to the characteristics of the three-phase AC curve. Therefore, by determining whether the data and values ​​at each sampling time are close to 0, data consistency judgment can be performed quickly, improving the execution speed of the method.

[0084] If the sum of data and values ​​at at least one sampling moment is greater than or equal to a first preset threshold, the difference between the sum of data and values ​​at each sampling moment and the sum of data and values ​​at the previous sampling moment can be further calculated to obtain the difference of data and values ​​at each sampling moment. Then, based on the difference of data and values ​​at each sampling moment, a data consistency judgment is performed on the data and values ​​at each sampling moment. Here, each sampling moment includes all sampling moments within a predetermined time period except for the first sampling moment. According to this embodiment, the difference of data and values ​​at each sampling moment can be further used to perform data consistency judgment, thereby improving data availability and ensuring the reliability of data consistency judgment.

[0085] Alternatively, it can be determined that the sum of data and values ​​at each sampling time is close to 0, and instead, the difference between the sum of data and values ​​at each sampling time can be used directly for data consistency judgment. That is, in step S903, the difference between the sum of data and values ​​at each sampling time and the sum of data and values ​​at the previous sampling time can be calculated to obtain the difference between the sum of data and values ​​at each sampling time; then, based on the difference between the sum of data and values ​​at each sampling time, a data consistency judgment is performed on the sum of data and values ​​at each sampling time. As mentioned above, each sampling time includes all sampling times within a predetermined time period except for the first sampling time. This avoids repeatedly performing data consistency judgments and improves the method execution speed.

[0086] According to one embodiment of this disclosure, the step of determining data consistency based on the difference between data and values ​​at each sampling time may include: for each sampling time, determining whether the absolute value of the difference between the data and values ​​at the corresponding sampling time is less than a second preset threshold, and in response to the absolute value of the difference between the data and values ​​at the corresponding sampling time being less than the second preset threshold, accumulating a first count value; and in response to the first count value being greater than the first count threshold, determining that the data and values ​​at each sampling time are consistent. Here, the first count value can be reset to zero when the anomaly detection method of the pitch system is started. The second preset threshold can be in the range of 5 to 10, but this disclosure is not limited thereto. In addition, the first count threshold can be set to an integer value greater than or equal to 10, but this disclosure is not limited thereto. For example, the first count threshold can also be set to an integer value greater than or equal to 5. In fact, the setting of the first count threshold needs to be considered from two aspects. On the one hand, the first count threshold can be set to a larger value, so that the accuracy of anomaly identification will be higher. On the other hand, the first count threshold cannot be set too large, otherwise it will lead to missed detection. According to equation (2), since the deviation of wind direction may have a certain impact on the centerline of torque value, when judging data consistency, it is possible to judge whether the difference between the data and values ​​at the sampling time before and after is close (i.e., whether the change of values ​​is stable), thereby realizing the data consistency judgment. In this way, the accuracy and reliability of data consistency judgment can be improved.

[0087] Table 2 shows the comparison with Figure 6 The table shows the data and value differences at certain sampling times corresponding to the torque curves of the three pitch motors during the pitch adjustment of the wind turbine generator set.

[0088] Table 2

[0089]

[0090]

[0091]

[0092] As shown in Table 2, the maximum difference between the data and values ​​at each sampling time is -4.6, and their absolute values ​​are all less than 5. Therefore, it can be determined that the data and values ​​at each sampling time are consistent.

[0093] Optionally, the step of determining data consistency based on the difference between data and values ​​at each sampling time includes: for each sampling time, determining whether the absolute value of the difference between the data and values ​​at the corresponding sampling time is greater than or equal to a second preset threshold; and in response to the absolute value of the difference between the data and values ​​at the corresponding sampling time being greater than or equal to the second preset threshold, accumulating a second count value; and in response to the second count value being greater than the second count threshold, determining that the data and values ​​at each sampling time are not consistent. Here, the second count value can be reset to zero when the anomaly detection method of the pitch system is started. As mentioned above, the second preset threshold can be in the range of 5 to 10, but this disclosure is not limited thereto. Furthermore, the second count threshold can be set to be the same as the first count threshold, or it can be set to be different from the first count threshold. In this way, the accuracy and reliability of data consistency determination can be improved.

[0094] When the data and values ​​at each sampling time are determined to be inconsistent as described above, in step S905, in response to the second count value being greater than the second count threshold, it can be determined that the abnormality of the pitch system is caused by an abnormality in the pitch bearing. This allows for more accurate detection of pitch bearing abnormalities (e.g., pitch bearing jamming), thereby improving the usability of the pitch system abnormality detection method.

[0095] Table 3 shows the comparison with Figure 7 The table shows the data and value differences at certain sampling times corresponding to the torque curves of the three pitch motors of the wind turbine generator set during operation and shutdown.

[0096] Table 3

[0097]

[0098]

[0099]

[0100]

[0101] As shown in Table 3, the absolute value of the difference between the data and the value is greater than the second preset threshold multiple times, which causes the second count value to be greater than the second count threshold. Therefore, it can be determined that the data and the value at each sampling time are not consistent, that is, there is an anomaly in the pitch system.

[0102] According to another embodiment of this disclosure, the step of determining data consistency for data and values ​​at each sampling time based on the difference between data and values ​​at each sampling time may include: for each sampling time, determining whether the sign of the difference between the data and values ​​at the corresponding sampling time and the second preset threshold is different from the sign of the difference between the data and values ​​at the previous sampling time and the second preset threshold; and in response to the difference between the sign of the difference between the data and values ​​at the corresponding sampling time and the second preset threshold being different from the sign of the difference between the data and values ​​at the previous sampling time and the second preset threshold, accumulating a third count value; and in response to the third count value being greater than the third count threshold, determining that the data and values ​​at each sampling time are not consistent. Here, the third count value can be reset to zero when the anomaly detection method of the pitch system is started. The third count threshold can be set to an integer value greater than or equal to 10, but this disclosure is not limited to this. For example, the third count threshold can also be set to an integer value greater than or equal to 5. In fact, the setting of the third count threshold also needs to be considered from two aspects. On the one hand, the third count threshold can be set to a larger value, so that the accuracy of anomaly identification will be higher. On the other hand, the third count threshold cannot be set too large, otherwise it will lead to missed detection. According to embodiments of this disclosure, if the difference between the data and the value at the sampling time crosses between being greater than or less than a second preset threshold (i.e., the difference between the data and the value and the second preset threshold is sometimes positive and sometimes negative), it indicates that the blades of the wind turbine generator may be malfunctioning. Specific physical phenomena can be described as follows: Figure 7 As shown. Therefore, by further judging the positive or negative change of the difference between the data and values ​​at each sampling time and the second preset threshold, the consistency of the data and values ​​at each sampling time can be determined, thereby improving the usability of the data.

[0103] Specifically, when it is determined above that the data and values ​​at each sampling moment are inconsistent, in step S905, in response to the inconsistency of the data and values ​​at each sampling moment, it is determined whether the sampling moments for accumulating the third count value are regularly distributed in time; in response to the regular distribution in time of the sampling moments for accumulating the third count value, it is determined that the abnormality of the pitch system is caused by the blade abnormality of the wind turbine generator; in response to the irregular distribution in time of the sampling moments for accumulating the third count value, it is determined that the abnormality of the pitch system is caused by the pitch bearing abnormality.

[0104] The principle for determining whether a pitch system malfunction is caused by a pitch bearing abnormality or a wind turbine blade abnormality is as follows. First, the force exerted by the blades on the pitch motor varies with the rotor rotation and is related to the blade azimuth angle. Therefore, its variation period is relatively long and exhibits a strong regularity with the rotor speed. Thus, if the difference between the data and the value and the difference from the second preset threshold shows a relatively regular positive and negative change, such as... Figure 7As shown, this allows for further assessment of blade anomalies. Secondly, due to the ball bearing structure of the pitch bearing, an anomaly in the pitch bearing intermittently affects the pitch motor current. For the pitch drive: as the load continuously increases, it affects the main magnetic flux, i.e., armature reaction. In this situation, the input current also increases to offset the load-induced reaction, keeping the total magnetic flux essentially constant. When the pitch bearing malfunctions, the load on the pitch bearing fluctuates during blade rotation, leading to significant current fluctuations, which are unrelated to the impeller speed. Therefore, if the difference between the data and the value does not exhibit a regular positive or negative change compared to the second preset threshold, a pitch bearing anomaly can be further identified. This allows for more accurate detection of both pitch bearing and blade anomalies, thereby improving the usability of the anomaly detection method for the pitch system.

[0105] Figure 10 This is a graph showing the current values ​​of the three pitch motors when the pitch bearing malfunctions. Figure 10 In the graph, the horizontal axis represents the time value, with time 0 indicating the moment when the wind turbine generator triggers a fault shutdown. The vertical axis represents the relevant variable values. Curve 1001 shows the operating data (current) of pitch motor 1, curve 1002 shows the operating data (current) of pitch motor 2, and curve 1003 shows the operating data (current) of pitch motor 3.

[0106] right Figure 10 The operating data of the three pitch motors are summed at each sampling time, and points are taken at equal intervals to obtain the results. Figure 11 The waveform diagram shows the data and values ​​at each sampling time. From Figure 11 As can be seen, the fluctuations in data and values ​​are completely irregular, and this characteristic can serve as a basis for distinguishing the types of anomalies in the pitch system.

[0107] Figure 12 This is a block diagram illustrating a computing device according to an embodiment of the present disclosure. As described above, the computing device may be a SCADA industrial control computer arranged in a wind farm control center, but the present disclosure is not limited thereto.

[0108] Reference Figure 12The computing device 1200 according to embodiments of the present disclosure includes a processor 1210 and a memory 1220. The processor 1210 may include (but is not limited to) a central processing unit (CPU), a digital signal processor (DSP), a microcomputer, a field-programmable gate array (FPGA), a system-on-a-chip (SoC), a microprocessor, an application-specific integrated circuit (ASIC), etc. The processor 1210 can acquire SCADA operating data of a target wind turbine generator. The memory 1220 can store computer programs to be executed by the processor 1210. The memory 1220 may include high-speed random access memory and / or a non-volatile computer-readable storage medium. When the processor 1210 executes the computer program stored in the memory 1220, the anomaly detection method for the pitch system as described above can be implemented.

[0109] The anomaly detection method for a pitch system according to embodiments of this disclosure can be programmed into a computer program and stored on a storage medium. When the computer program is executed by a processor, the anomaly detection method for a pitch system as described above can be implemented. Examples of storage media include: read-only memory (ROM), random access programmable read-only memory (PROM), electrically erasable programmable read-only memory (EEPROM), random access memory (RAM), dynamic random access memory (DRAM), static random access memory (SRAM), flash memory, non-volatile memory, CD-ROM, CD-R, CD+R, CD-RW, CD+RW, DVD-ROM, DVD-R, DVD+R, DVD-RW, DVD+RW, DVD-RAM, BD-ROM, BD-R, BD-R LTH, BD-RE, Blu-ray or optical disc storage, hard disk drive (HDD), solid-state drive (SSD), card storage (such as multimedia cards, secure digital (SD) cards, or ultra-fast digital (XD) cards), magnetic tape, floppy disk, magneto-optical data storage device, optical data storage device, hard disk, solid-state drive, and any other device configured to store a computer program and any associated data, data files, and data structures in a non-transitory manner and to provide the computer program and any associated data, data files, and data structures to a processor or computer so that the processor or computer can execute the computer program. In one example, the computer program and any associated data, data files, and data structures are distributed across a networked computer system, such that the computer program and any associated data, data files, and data structures are stored, accessed, and executed in a distributed manner through one or more processors or computers.

[0110] The anomaly detection method, storage medium, and computing device of the pitch system according to the embodiments of this disclosure can realize anomaly diagnosis and early warning based on SCADA operation data, effectively solve the problem of excessively long SCADA operation data sampling period, realize the correct parsing of wind turbine operation data that cannot be achieved by existing spectrum analysis, jump detection, amplitude detection, etc., and improve the application value of operation data.

[0111] Furthermore, the anomaly detection method, storage medium, and computing device of the pitch system according to the embodiments of this disclosure can diagnose and analyze bearing anomalies and blade anomalies of the pitch system by performing anomaly detection based on SCADA operating data, and the detection accuracy is high and is not affected by factors such as wind speed, wind direction, and ambient temperature.

[0112] Furthermore, the anomaly detection method, storage medium, and computing device for the pitch system according to the embodiments of this disclosure do not require upgrading or optimizing the PLC program, nor do they require repeated adjustments to the detection parameters. They are particularly suitable for SCADA transient data files generated from historical dates, greatly improving the effectiveness and reliability of anomaly detection.

[0113] Furthermore, the anomaly detection method, storage medium, and computing device for the pitch system according to embodiments of this disclosure are minimally affected by torque value fluctuations because they do not involve the detection of spectrum or amplitude jumps, effectively reducing the probability of false detections. Simultaneously, they solve the problem of a small number of B files (generated only during shutdown), enabling 24-hour monitoring of the pitch system's operating status; and they do not require modification of SCADA programs or communication protocols between the wind turbine generator and central monitoring, thus facilitating implementation and reducing development workload.

[0114] Furthermore, the anomaly detection method, storage medium, and computing device for the pitch system according to embodiments of this disclosure do not require recalculation of operating data based on wind direction values ​​in practical applications. Instead, they process the operating data according to its characteristics, thus improving the accuracy of anomaly detection. This is because wind direction changes instantaneously. Although the force on the blades is related to wind direction, incorporating wind direction data into the calculation would make the data cluttered and generate significant interference.

[0115] While some embodiments of this disclosure have been shown and described, those skilled in the art will understand that modifications may be made to these embodiments without departing from the principles and spirit of this disclosure, which are defined by the claims and their equivalents.

Claims

1. A method for detecting anomalies in a pitch system, characterized in that, The anomaly detection method includes: During the pitch control operation, the operating data of each pitch motor of the pitch control system is acquired. The operating data of each pitch motor at each sampling time within a predetermined time period are summed to obtain the data and value at each sampling time. Perform a data consistency check on the data and values ​​at each sampling time; The consistency of data and values ​​at each sampling time confirms that the pitch system is operating normally; and The inconsistency between the data and values ​​at each sampling time indicates an anomaly in the pitch system.

2. The anomaly detection method as described in claim 1, characterized in that, In response to inconsistencies in data and values ​​at each sampling time, the steps to determine an anomaly in the pitch system include: Based on the data and values ​​at each sampling time, determine whether the anomaly in the pitch system is caused by a pitch bearing malfunction or a blade malfunction in the wind turbine generator.

3. The anomaly detection method as described in claim 1, characterized in that, The anomaly detection method further includes: During the pitch control operation of the pitch system, the rotor speed of the wind turbine generator is obtained; Calculate the impeller rotation period based on the impeller speed; In response to the time inconsistency between the impeller rotation period and the sampling period, the step of summing up the operating data of each pitch motor at each sampling moment within a predetermined time period is performed.

4. The anomaly detection method as described in claim 1, characterized in that, The steps for determining data consistency for each sampling time point include: Determine whether the data and values ​​at each sampling time are all less than a first preset threshold; In response to the fact that the data and values ​​at each sampling time are all less than the first preset threshold, it is determined that the data and values ​​at each sampling time are consistent.

5. The anomaly detection method as described in claim 4, characterized in that, The steps for determining data consistency for each sampling time point also include: In response to the fact that the sum of data and values ​​at at least one sampling time is greater than or equal to the first preset threshold, the difference between the sum of data and values ​​at each sampling time and the sum of data and values ​​at the previous sampling time is calculated to obtain the difference between the sum of data and values ​​at each sampling time, wherein each sampling time includes all sampling times other than the first sampling time within a predetermined time period; Based on the difference between the data and values ​​at each sampling time, a data consistency judgment is made on the data and values ​​at each sampling time.

6. The anomaly detection method as described in claim 1, characterized in that, The steps for determining data consistency for each sampling time point include: Calculate the difference between the sum of data and values ​​at each sampling time and the sum of data and values ​​at the previous sampling time to obtain the difference between the sum of data and values ​​at each sampling time. Each sampling time includes all sampling times within a predetermined time period except for the first sampling time. Based on the difference between the data and values ​​at each sampling time, a data consistency judgment is made on the data and values ​​at each sampling time.

7. The anomaly detection method as described in claim 5 or 6, characterized in that, The steps for determining data consistency based on the difference between the data and values ​​at each sampling time include: For each sampling time, determine whether the absolute value of the difference between the data and the value at the corresponding sampling time is less than a second preset threshold, and in response to the absolute value of the difference between the data and the value at the corresponding sampling time being less than the second preset threshold, accumulate the first count value; In response to a first count value being greater than a first count threshold, it is determined that the data and values ​​at each sampling time are consistent.

8. The anomaly detection method as described in claim 5 or 6, characterized in that, The steps for determining data consistency based on the difference between the data and values ​​at each sampling time include: For each sampling time, determine whether the absolute value of the difference between the data and the value at the corresponding sampling time is greater than or equal to the second preset threshold, and in response to the absolute value of the difference between the data and the value at the corresponding sampling time being greater than or equal to the second preset threshold, accumulate the second count value; In response to a second count value being greater than a second count threshold, it is determined that the data and values ​​at each sampling time are inconsistent.

9. The anomaly detection method as described in claim 8, characterized in that, In response to inconsistencies in data and values ​​at each sampling time, the steps to determine an anomaly in the pitch system include: In response to the second count value being greater than the second count threshold, it is determined that the abnormality in the pitch system is caused by a pitch bearing malfunction.

10. The anomaly detection method as described in claim 5 or 6, characterized in that, The steps for determining data consistency based on the difference between the data and values ​​at each sampling time include: For each sampling time, determine whether the sign of the difference between the data and values ​​at the corresponding sampling time and the second preset threshold is different from the sign of the difference between the data and values ​​at the previous sampling time and the second preset threshold. In response to the difference between the sign of the difference between the data and values ​​at the corresponding sampling time and the second preset threshold and the sign of the difference between the data and values ​​at the previous sampling time and the second preset threshold, accumulate the third count value. In response to a third count value being greater than a third count threshold, it is determined that the data and values ​​at each sampling time are inconsistent.

11. The anomaly detection method as described in claim 10, characterized in that, In response to inconsistencies in data and values ​​at each sampling time, the steps to determine an anomaly in the pitch system include: In response to the inconsistency of data and values ​​at each sampling time, determine whether the sampling times for accumulating the third count value are regularly distributed over time; In response to the regular temporal distribution of the sampling times at which the third count value is accumulated, it is determined that the anomaly in the pitch system is caused by a blade anomaly in the wind turbine generator set. In response to the irregular temporal distribution of the sampling times at which the third count value is accumulated, it is determined that the anomaly in the pitch system is caused by a pitch bearing malfunction.

12. The anomaly detection method as described in claim 1, characterized in that, The operating data for each pitch motor is SCADA operating data, and includes at least one of the following: torque value, current value, given speed value, and actual speed value for each pitch motor.

13. A storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the anomaly detection method for the pitch system as described in any one of claims 1 to 12.

14. A computing device, characterized in that, The computing device includes: processor; and A memory storing a computer program that, when executed by a processor, implements the anomaly detection method for a pitch system as described in any one of claims 1 to 13.

Images