Fault online identification method for ensuring safety of wind turbine generator

Abstract

The invention provides a fault online identification method for ensuring the safety of a wind turbine generator, and relates to the technical field of industrial detection and automation. The method comprises the following steps: collecting vibration data and rotating speed data of a target bearing; determining a rotation period based on the data acquisition interval and the rotation speed data at each sampling moment; determining the abnormal probability of the rotation period based on the rotation speed mean value and the rotation speed change total value; determining the rotation period of which the abnormal probability is greater than a preset threshold value as an abnormal rotation period, and determining a similar rotation period of the abnormal rotation period in the normal rotation period; determining a filtering window of the abnormal rotation period; determining matching data of the vibration data in the abnormal rotation period at the corresponding position of the similar rotation period, and generating a matching window; determining the filtering weight of the matched window, and updating the vibration data in the abnormal rotation period according to the filtering weight; and identifying whether the target bearing has a fault based on the updated vibration data in each rotation period. According to the invention, the wind turbine generator bearing fault identification accuracy is improved.

Description

Technical Field

[0001] This application relates to the field of industrial testing and automation technology, specifically to an online fault identification method for ensuring the safety of wind turbine units. Background Technology

[0002] As a crucial component of clean energy, the safety of wind turbines directly impacts power generation efficiency and long-term stable operation. Therefore, fault identification in wind turbines is a critical aspect of this field. Real-time vibration data of wind turbine bearings during continuous rotation exhibits distinct fault characteristics and is thus frequently used for fault identification.

[0003] In existing technologies, the periodicity of bearing vibration data fluctuations is typically analyzed based on the periodic characteristics of the bearing's circumferential rolling motion to identify whether a bearing has failed. However, dynamic changes in wind speed during actual operation can alter bearing rotation speed, causing changes in the periodicity of bearing vibration and thus affecting the accuracy of bearing fault identification. Furthermore, wind speed variations can also cause bearing vibrations, which may mask the fault fluctuations caused by bearing malfunctions, further impacting the accuracy of bearing fault identification. Summary of the Invention

[0004] In view of this, this application proposes an online fault identification method to ensure the safety of wind turbine units, in order to solve the problem in related technologies that the periodic performance of bearing vibration changes due to dynamic changes in wind speed, and the bearing vibration caused by wind speed may cover the fault fluctuations caused by bearing faults, thereby affecting the accuracy of bearing fault identification.

[0005] According to the first aspect of this application, an online fault identification method for ensuring the safety of wind turbine generators is provided, and the specific technical solution adopted is as follows: Collect vibration and rotational speed data of the target bearing in the wind turbine unit to be identified; The rotation period of the target bearing is determined based on the data acquisition interval and the rotational speed data at each sampling time. For each rotation cycle, the mean rotation speed of the rotation speed data and the total value of the rotation speed change at each sampling time when the rotation speed data changes are obtained, and the anomaly probability is determined based on the mean rotation speed and the total value of the rotation speed change; The rotation cycle with an abnormal probability greater than a first preset threshold is defined as an abnormal rotation cycle, the other rotation cycles are defined as normal rotation cycles, and similar rotation cycles to the abnormal rotation cycle are determined among the normal rotation cycles. The filter window for the abnormal rotation period is determined based on the abnormal probability and the preset initial window. Based on the rotation angle of the target bearing relative to the start time of the cycle at the sampling time corresponding to each vibration data within the abnormal rotation cycle, the matching data of the vibration data at the corresponding position in the similar rotation cycle is determined, and a matching window is generated; Determine the filtering weight of each matching window, and filter the abnormal rotation cycle within the filtering window through the matching window based on the filtering weight, and update the vibration data in the abnormal rotation cycle; The target bearing is identified as having failed based on the updated vibration data within each rotation cycle.

[0006] For example, determining the rotation period of the target bearing based on the data acquisition interval and the rotational speed data at each sampling moment includes: setting the initial rotation angle of the target bearing to 0; calculating the product of the rotational speed data at the corresponding sampling moment and the data acquisition interval at each data acquisition interval to obtain the cumulative value of the rotational angle at each sampling moment relative to the previous sampling moment; sequentially adding the cumulative value of the rotational angle to the initial rotational angle to obtain the current rotational angle corresponding to each sampling moment; when the current rotational angle reaches 360 degrees, determining the time interval formed by the current sampling moment and the sampling moment corresponding to the initial rotational angle as the rotation period; clearing the current rotational angle to zero and recording the cleared current rotational angle as the initial rotational angle; repeating the process of calculating the product of the rotational speed data at the corresponding sampling moment and the data acquisition interval at each data acquisition interval and recording the cleared current rotational angle as the initial rotational angle, until the entire rotation period of the target bearing up to the current moment is determined.

[0007] For example, the step of obtaining the mean rotational speed of the rotational speed data and the total rotational speed change at each sampling moment when the rotational speed data changes for each rotational cycle, and determining the anomaly probability based on the mean rotational speed and the total rotational speed change, includes: obtaining the rotational speed data of the target bearing at each sampling moment within the rotational cycle and calculating its average value to obtain the mean rotational speed; obtaining the sampling moment when the rotational speed data of the target bearing changes, and recording it as the rotational speed change moment; determining the rotational speed change value of each rotational speed change moment relative to its previous sampling moment and summing them to obtain the total rotational speed change; calculating the product of the mean rotational speed and the total rotational speed change and performing normalization processing to obtain the anomaly probability of the rotational cycle.

[0008] For example, determining a similar rotation cycle of the abnormal rotation cycle in the normal rotation cycle includes: calculating the cycle length difference between the abnormal rotation cycle and each normal rotation cycle for each abnormal rotation cycle; and determining the normal rotation cycle corresponding to the cycle length difference that is less than a second preset threshold as the similar rotation cycle of the abnormal rotation cycle.

[0009] For example, determining the filtering window for the abnormal rotation cycle based on the abnormal probability and a preset initial window includes: setting a preset initial window, calculating the product of the window size of the preset initial window and the abnormal probability and rounding it up, denoted as the window adjustment factor; if the window adjustment factor is even, determining the window adjustment factor as the window increment; if the window adjustment factor is odd, determining the sum of the window adjustment factor plus 1 as the window increment; and determining the filtering window based on the preset initial window and the window increment.

[0010] For example, determining the matching data of the vibration data at the corresponding position in the similar rotation cycle based on the rotation angle of the target bearing relative to the start time of the cycle at the sampling time corresponding to each vibration data in the abnormal rotation cycle, and generating a matching window, includes: for each vibration data in the abnormal rotation cycle, determining the first sampling time corresponding to the vibration data in the abnormal rotation cycle; within a first time interval from the start time of the cycle to the first sampling time in the abnormal rotation cycle, calculating and accumulating the product of the rotational speed data at the corresponding sampling time and the data acquisition interval at each data acquisition interval to obtain the first rotation angle of the vibration data in the abnormal rotation cycle; determining the second sampling time corresponding to the first rotation angle in each similar rotation cycle, and recording the vibration data corresponding to the second sampling time as the matching data; and constructing the matching window based on the matching data.

[0011] For example, determining the filtering weight of each matching window includes: determining the average value of the rotational speed data of the target bearing within the filtering window, denoted as a first rotational speed factor; determining the average value of the rotational speed data of the target bearing within the matching window, denoted as a second rotational speed factor; determining the absolute value of the difference between the first rotational speed factor and the second rotational speed factor, denoted as a first filtering factor; calculating the difference in period length between the abnormal rotational period corresponding to the filtering window and the similar rotational period corresponding to the matching window, denoted as a second filtering factor; determining the filtering weight factor corresponding to the matching window based on the first filtering factor and the second filtering factor, and calculating the sum of the filtering weight factors corresponding to each matching window; and for each matching window, determining the ratio of the filtering weight factor corresponding to the matching window to the sum of the filtering weight factors as the filtering weight of the matching window.

[0012] For example, the step of filtering the abnormal rotation cycle through the matching window based on the filtering weight and updating the vibration data in the abnormal rotation cycle includes: performing a weighted operation on the vibration data in the matching window based on the filtering weight to obtain a filtering result, and using the filtering result to update the vibration data in the abnormal rotation cycle.

[0013] For example, the method further includes: when there are multiple filtering windows in the abnormal rotation cycle, calculating the average value of the filtering results corresponding to each filtering window, and updating the vibration data in the abnormal rotation cycle with the average value of the filtering results.

[0014] For example, identifying whether the target bearing has failed based on the updated vibration data within each rotation cycle includes: for each vibration data within each rotation cycle, calculating the average value of the vibration data other than the current vibration data within the rotation cycle, and calculating the absolute value of the difference between the average value and the current vibration data, denoted as a fault factor; when the fault factor is greater than a third preset threshold, determining the first matching data of the current vibration data in the previous rotation cycle and the second matching data in the next rotation cycle; if the fault factors corresponding to the first matching data and the second matching data are both greater than the third preset threshold, determining that the target bearing has failed, and issuing a fault warning.

[0015] This application may have some or all of the following beneficial effects: In the online fault identification method for ensuring the safety of wind turbine units provided in this application, the rotation period of the target bearing is determined based on the data acquisition interval and the rotational speed data at each sampling moment. By dynamically tracking the rotational speed data at each sampling moment and combining it with the data acquisition interval to determine the rotation period, the accuracy of the target bearing rotation period division is improved, thus ensuring that subsequent vibration data analysis is based on accurate rotation period division. By determining similar rotation periods of abnormal rotation periods within normal rotation periods and generating a matching window based on the matching data of each vibration data point within the abnormal rotation period at the corresponding position of the similar rotation period, the vibration data within the abnormal rotation period can be filtered using the vibration data of similar rotation periods. This smooths out normal vibrations caused by changes in the target bearing rotational speed, preventing them from masking abnormal vibrations caused by bearing faults, and making fault characteristics easier to identify. By determining different filtering weights for the matching window and combining these weights to filter abnormal rotation periods within the filtering window, the reliability of the updated vibration data is improved. Based on the updated data, the accuracy of bearing fault identification is effectively improved, thereby ensuring the safe and stable operation of the wind turbine unit.

[0016] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and do not limit this application. Attached Figure Description

[0017] To more clearly illustrate the technical solutions and advantages in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 A flowchart is shown for an online fault identification method to ensure the safety of wind turbine units according to an exemplary embodiment of this application. Detailed Implementation

[0019] To further illustrate the technical means and effects adopted by this application to achieve the intended purpose of the invention, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of an online fault identification method for ensuring the safety of wind turbine units proposed in this application. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0020] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0021] The following description, in conjunction with the accompanying drawings, details the specific scheme of the online fault identification method for ensuring the safety of wind turbine units provided in this application.

[0022] Please see Figure 1 The document illustrates a flowchart of an online fault identification method for ensuring the safety of wind turbine generators, as provided in one embodiment of this application. Figure 1 As shown, the online fault identification method for ensuring the safety of wind turbine units specifically includes the following steps: S110: Collect vibration and rotational speed data of the target bearing in the wind turbine unit to be identified; S120: Determine the rotation period of the target bearing based on the data acquisition interval and the rotational speed data at each sampling time; S130: For each rotation cycle, obtain the mean speed of the speed data and the total speed change at each sampling time when the speed data changes, and determine the anomaly probability based on the mean speed and the total speed change. S140: Determine the rotation cycle with an abnormal probability greater than the first preset threshold as the abnormal rotation cycle, determine the other rotation cycles as normal rotation cycles, and determine the similar rotation cycles of the abnormal rotation cycle in the normal rotation cycles. S150: A filter window that determines the abnormal rotation period based on the abnormal probability and a preset initial window; S160: Based on the rotation angle of the target bearing relative to the start time of the cycle at the sampling time corresponding to each vibration data within the abnormal rotation cycle, determine the matching data of the vibration data at the corresponding position of similar rotation cycles, and generate a matching window; S170: Determine the filtering weight of each matching window, and filter the abnormal rotation cycle within the filtering window through the matching window based on the filtering weight, and update the vibration data in the abnormal rotation cycle. S180: Identify whether the target bearing has failed based on the updated vibration data within each rotation cycle.

[0023] The following is a detailed explanation of each step in the above-mentioned online fault identification method for ensuring the safety of wind turbine units: In step S110, vibration data and rotational speed data of the target bearing in the wind turbine to be identified are collected.

[0024] In this embodiment, the wind turbine to be identified refers to a wind turbine that requires online fault identification. Since the vibration data of the bearings in a wind turbine during continuous rotation exhibits obvious fault characteristics, the method provided in this embodiment identifies the faults in the wind turbine by determining whether a fault exists in the target bearing.

[0025] In this embodiment, the target bearing is the bearing in the wind turbine unit to be identified that requires key monitoring for faults. This target bearing can be set based on the actual scenario. For example, in a specific implementation, the target bearing can be the high-speed bearing of the gearbox in the wind turbine unit to be identified. Its fault has a significant impact on the safety of the wind turbine unit, and abnormal vibration during rotation is the core manifestation of the fault.

[0026] In this embodiment, the vibration data refers to the vibration signal of the target bearing during rotation, reflecting the vibration state of the bearing. For example, this vibration data can be acquired using an accelerometer.

[0027] In this embodiment, the aforementioned rotational speed data reflects the rotational speed of the target bearing. For example, this rotational speed data can be acquired by a rotational speed sensor, including the instantaneous rotational speed of the target bearing at each sampling moment.

[0028] In one specific implementation of this application embodiment, the acquisition of vibration and rotational speed data of the target bearing in the wind turbine to be identified can be achieved as follows: Vibration data of the target bearing is acquired using an accelerometer installed on the target bearing housing or a rigid structure near the target bearing. This accelerometer can be bolted on to reduce resonance interference and is suitable for long-term acquisition. Its parameters can be set as follows: measurement range set to ±50g to cover normal vibration and fault impact; frequency response range set to 10Hz-10kHz to cover low-frequency rotational vibration and high-frequency fault impact; sensitivity set to 500mV / g to capture weak vibration signals. Rotational speed data of the target bearing is acquired using a magnetoelectric rotational speed sensor installed near the gear or magnet at the end of the target shaft. The sensor outputs one or more pulse signals per revolution to mark the rotation cycle. It can be installed non-contactly with a gap of 0.5-2mm to avoid wear caused by contact friction between the magnetoelectric speed sensor and the target bearing, ensuring long-term stable data acquisition. Its parameter settings can be as follows: measurement range 0-3000 r / min (this parameter must be adapted to the target bearing's speed range); vibration data and speed data are synchronously acquired via an external clock on the acquisition card, ensuring that the timestamp error is less than a preset value (e.g., 1 μs). The sampling frequency of the acquisition card is set to at least twice the highest frequency component in the vibration signal caused by the target bearing fault, according to the sampling theorem. The highest fault frequency in the vibration signal caused by the bearing fault is 8 kHz; therefore, exemplarily, the sampling frequency of the acquisition card can be set to 51.2 kHz. It should be noted that the above scenario is only an illustrative example, and the scope of protection of this application embodiment is not limited thereto. For example, other data acquisition devices that can achieve the same function can also be used in this application embodiment, or the parameters can be set to other values ​​based on actual needs.

[0029] In step S120, the rotation period of the target bearing is determined based on the data acquisition interval and the rotational speed data at each sampling time.

[0030] In the embodiments of this application, the above sampling time refers to the specific time point at which vibration data and rotational speed data are collected, and each sampling time corresponds to a set of vibration data and rotational speed data of the target bearing.

[0031] In this embodiment, the data acquisition interval refers to the time interval between two adjacent data samples, which is determined by the sampling frequency set above. For example, if the sampling frequency is 51.2kHz, then the data acquisition interval is (1 / 51200) seconds.

[0032] In the embodiments of this application, the aforementioned rotation period refers to the time required for the target bearing to rotate one revolution (360 degrees), which is used to analyze the periodic characteristics of vibration data. Its period length changes dynamically with the rotation speed of the target bearing. The higher the rotation speed of the target bearing, the shorter the period length of the rotation period.

[0033] For example, the above method for determining the rotation period of the target bearing based on the data acquisition interval and the rotational speed data at each sampling moment can be implemented as follows: Set the initial rotation angle of the target bearing to 0. At each data acquisition interval, calculate the product of the rotational speed data at the corresponding sampling moment and the data acquisition interval to obtain the cumulative value of the rotational angle at each sampling moment relative to the previous sampling moment. Accumulate the cumulative value of the rotational angle to the initial rotational angle to obtain the current rotational angle corresponding to each sampling moment. When the current rotational angle reaches 360 degrees, determine the time interval formed by the current sampling moment and the sampling moment corresponding to the initial rotational angle as the rotation period. Clear the current rotational angle to zero and record the cleared current rotational angle as the initial rotational angle. Repeat the process of "calculating the product of the rotational speed data at the corresponding sampling moment and the data acquisition interval at each data acquisition interval" to "recording the cleared current rotational angle as the initial rotational angle" until the entire rotation period of the target bearing up to the current moment is determined.

[0034] In one specific implementation of this application embodiment, the determination of the rotation period of the target bearing can be achieved as follows: starting from the start time of fault identification, the rotational speed data of the target bearing at each sampling time is recorded in real time, and a correspondence between the sampling time and the rotational speed data is established; taking the first sampling time after the start time of fault identification as an example, the initial rotational angle corresponding to the start time of fault identification is 0. After a data acquisition interval, the instantaneous rotational speed corresponding to the first sampling time is obtained, and the product of the instantaneous rotational speed data and the data acquisition interval is calculated to obtain the cumulative value of the rotational angle from the start time of fault identification to the first sampling time. This cumulative value of the rotational angle is added to the initial rotational angle to obtain the current rotational angle corresponding to the first sampling time; after another... After one data acquisition interval, the instantaneous rotational speed corresponding to the second sampling moment is obtained. The product of this instantaneous rotational speed and the data acquisition interval is calculated to obtain the cumulative value of the rotation angle from the first sampling moment to the second sampling moment. This cumulative value of the rotation angle is added to the current rotation angle corresponding to the first sampling moment to obtain the current rotation angle corresponding to the second sampling moment. This process is repeated until the current rotation angle corresponding to a certain sampling moment is 360 degrees. The time interval between this sampling moment and the fault identification start moment is then determined as the first rotation cycle. The current rotation angle is then cleared to zero, and the second rotation cycle is determined in the same way based on the method used to determine the first rotation cycle. This process is repeated to determine all rotation cycles of the target bearing from the fault identification start moment to the current moment.

[0035] In step S130, for each rotation cycle, the mean rotation speed of the rotation speed data and the total value of the rotation speed change at each sampling time when the rotation speed data changes are obtained, and the anomaly probability is determined based on the mean rotation speed and the total value of the rotation speed change.

[0036] The failure of the target bearing typically manifests as fluctuations in vibration data. However, dynamic changes in rotational speed can also cause vibration in the target bearing, which may mask the vibration caused by the bearing failure itself, thus affecting the accuracy of bearing failure identification. To address this issue, this application embodiment distinguishes between normal and abnormal rotational cycles within the rotational cycle. This allows for filtering of the vibration data during abnormal rotational cycles in subsequent processes, eliminating the impact of dynamic changes in rotational speed during these cycles.

[0037] In this embodiment, the aforementioned anomaly probability is an indicator used to distinguish between normal and abnormal rotation cycles. This anomaly probability measures the degree to which the rotation cycle is affected by the rotational speed; the higher the value, the higher the probability that the corresponding rotation cycle is abnormal due to the influence of the rotational speed.

[0038] For example, the anomaly probability can be determined based on the average speed and the total speed change within the rotation cycle. The average speed is the average value of the speed data corresponding to all sampling times within the rotation cycle, reflecting the average rotational speed of the target bearing within that rotation cycle; the total speed change is the sum of the speed changes corresponding to all sampling times where speed changes occur within the rotation cycle, measuring the total amplitude of speed fluctuations within that cycle.

[0039] The above-mentioned method, which acquires the average speed of the rotational speed data and the total speed change at each sampling moment when the speed data changes for each rotational cycle, can be used to determine the anomaly probability based on the average speed and the total speed change. This can be achieved as follows: Acquire the target bearing's speed data at each sampling moment within the rotational cycle and calculate its average value to obtain the average speed; acquire the sampling moments when the target bearing's speed data changes, and record them as speed change moments. For example, assuming a rotational cycle includes 5 sampling moments with corresponding speed data of 1000 r / s, 1000 r / s, 1200 r / s, 1200 r / s, and 1500 r / s, then the aforementioned speed change moments are the third and fifth sampling moments; determine the speed change value of each speed change moment relative to its previous sampling moment and sum them to obtain the total speed change value. For example, in the above rotation cycle, the rotation speed change value corresponding to the third sampling time is 200 r / s, the rotation speed change value corresponding to the fifth sampling time is 300 r / s, and the total rotation speed change value is 500 r / s; the product of the mean rotation speed and the total rotation speed change value is calculated and normalized to obtain the abnormal probability of the rotation cycle.

[0040] Specifically, after determining the average speed and the total speed variation, the above-mentioned anomaly probability can be calculated using the following formula: in, This represents the probability of an anomaly in the current rotation cycle. This represents the total change in rotational speed within the current rotational cycle. This represents the average rotational speed within the current rotational cycle. The normalization function; the average rotational speed during the current rotational cycle. The larger the value, the greater the likelihood that the vibration generated by the target bearing due to the rotational speed will cover the fault vibration, and the higher the probability of anomalies in the current rotational cycle. The higher the value, the better; the total value of the speed change in the current rotation cycle. The larger the value, the greater the change in the target bearing's rotational speed, the greater the possibility that the resulting vibration fluctuations will cover the fault vibrations, and the higher the abnormal probability P of the current rotation cycle.

[0041] In step S140, rotation cycles with an abnormal probability greater than a first preset threshold are defined as abnormal rotation cycles, other rotation cycles are defined as normal rotation cycles, and similar rotation cycles to abnormal rotation cycles are determined among normal rotation cycles.

[0042] In this embodiment, the aforementioned abnormal rotation cycle refers to a rotation cycle in which the target bearing rotation speed changes significantly and the average rotation speed is relatively high, making it highly likely that the vibration caused by the rotation speed will mask the vibration of the target bearing failure.

[0043] In the embodiments of this application, the above-mentioned normal rotation cycle refers to the rotation cycle in which the target bearing speed changes slowly and the average speed has a weak influence, and the vibration caused by the speed will not significantly mask the vibration of the target bearing fault.

[0044] In this embodiment of the application, the first preset threshold is a pre-set critical value used to distinguish between abnormal rotation cycles and normal rotation cycles. It is a standard for judging whether the rotation cycle is abnormal. The specific value can be set according to the actual operating conditions and historical data of the wind turbine.

[0045] For example, the above-mentioned determination of rotation cycles with an abnormal probability greater than a first preset threshold as abnormal rotation cycles and determination of other rotation cycles as normal rotation cycles can be achieved as follows: the first preset threshold is determined based on the abnormal probability distribution of historical normal operation data. For example, the first preset threshold is set to 1.2 times the maximum abnormal probability of historical normal cycles; all rotation cycles of the target bearing are traversed, the abnormal probability of each rotation cycle is compared with the first preset threshold, and rotation cycles with an abnormal probability greater than the first preset threshold are marked as abnormal rotation cycles, and other rotation cycles other than abnormal rotation cycles are marked as normal rotation cycles.

[0046] In the embodiments of this application, the aforementioned similar rotation cycle refers to a rotation cycle in which the difference in cycle length between the normal rotation cycle and the abnormal rotation cycle is small; since the cycle length is determined by the rotation speed, the closer the lengths of the similar rotation cycle and the abnormal rotation cycle are, the more similar their rotation speed characteristics are to those of the abnormal rotation cycle.

[0047] For example, the above-mentioned method of determining the similar rotation cycle of the abnormal rotation cycle in the normal rotation cycle can be implemented as follows: for each abnormal rotation cycle, calculate the difference in cycle length between the abnormal rotation cycle and each normal rotation cycle; determine the normal rotation cycle corresponding to the cycle length difference that is less than the second preset threshold as the similar rotation cycle of the abnormal rotation cycle.

[0048] In step S150, the filter window for the abnormal rotation period is determined based on the abnormal probability and the preset initial window.

[0049] During abnormal rotation cycles, the vibration caused by the target bearing's rotational speed can mask the abnormal vibration caused by the fault, thus affecting the identification of the target bearing fault. Since different rotation cycles correspond to the same rotation range of the target bearing, the embodiments of this application can filter the vibration data by using similar rotation cycles of the abnormal rotation cycle to smooth the vibration of the target bearing caused by the rotational speed during the abnormal rotation cycle, making it easier to identify fault vibration.

[0050] In this embodiment of the application, the aforementioned preset initial window is a pre-set reference size for the filter window, which is the initial value for calculating the abnormal rotation cycle filter window.

[0051] In this embodiment, the aforementioned filtering window refers to a sliding window that filters the vibration data within the abnormal rotation cycle. The size of the window is dynamically adjusted according to the abnormal probability of the abnormal rotation cycle. The larger the filtering window, the stronger the smoothing effect on the vibration fluctuations caused by the target bearing rotation speed. The smaller the filtering window, the more detailed features of the vibration data can be preserved.

[0052] When filtering abnormal rotation cycles, the larger the total value of the rotational speed change and the average rotational speed within the abnormal rotation cycle (i.e., the higher the probability of abnormality), the larger the required filtering window. For example, the filtering window for determining the abnormal rotation cycle based on the probability of abnormality and a preset initial window can be implemented as follows: Set a preset initial window, calculate the product of the window size and the probability of abnormality, and round it up; this is denoted as the window adjustment factor. If the window adjustment factor is even, it is determined as the window increment; if the window adjustment factor is odd, the sum of the window adjustment factor and 1 is determined as the window increment. The filtering window is then determined based on the preset initial window and the window increment.

[0053] Specifically, the above filtering window can be implemented as follows: Set a preset initial window with a window size of R0, for example, R0=5; calculate the window increment using the following formula: in, This is the window increment of the filtering window relative to the preset initial window; The probability of an anomaly in the current abnormal rotation cycle; Set as the default initial window; For window adjustment factor, It is a rounding function. The larger the value, the higher the probability of an anomaly in the current abnormal rotation cycle, and the larger the required filtering window and the corresponding window increment. Also bigger.

[0054] After determining the window increment, the size of the filtering window can be determined based on the preset initial window size and the window increment as follows: in, This refers to the size of the filtering window; Set the initial window size; This is the window increment of the filtering window relative to the preset initial window.

[0055] In addition, for vibration data near the edge within an abnormal rotation cycle, the corresponding filtering window may cover the data of adjacent rotation cycles. In this case, the covered data of adjacent cycles is used as temporary data to construct a complete filtering window, ensuring that each vibration data can be effectively filtered.

[0056] In step S160, the matching data of the vibration data at the corresponding position of similar rotation cycles is determined based on the rotation angle of the target bearing relative to the start time of the cycle at the sampling time corresponding to each vibration data within the abnormal rotation cycle, and a matching window is generated.

[0057] In this embodiment of the application, the aforementioned matching data refers to vibration data that has the same rotation angle relative to the starting time of a vibration data in an abnormal rotation cycle within a similar rotation cycle. Since the same rotation angle corresponds to the same position of the target bearing rotation, this serves as the matching basis, and the vibration data in the abnormal rotation cycle and its matching data in a similar rotation cycle have similar attributes.

[0058] In this embodiment, the matching window refers to a window composed of matching data within similar rotation cycles, and its size is consistent with the filtering window within abnormal rotation cycles, used to filter vibration data within abnormal rotation cycles.

[0059] For example, the above method of determining the matching data of vibration data at the corresponding position in similar rotation cycles based on the rotation angle of the target bearing relative to the start time of the cycle at the sampling time corresponding to each vibration data in an abnormal rotation cycle, and generating a matching window, can be achieved as follows: For each vibration data in an abnormal rotation cycle, determine the first sampling time corresponding to the vibration data in the abnormal rotation cycle; within the first time interval from the start time of the abnormal rotation cycle to the first sampling time, calculate and accumulate the product of the rotational speed data at the corresponding sampling time and the data acquisition interval at each data acquisition interval to obtain the first rotation angle of the vibration data in the abnormal rotation cycle; in each similar rotation cycle, determine the second sampling time corresponding to the first rotation angle in the similar rotation cycle, and record the vibration data corresponding to the second sampling time as the matching data; construct a matching window based on the matching data.

[0060] Specifically, taking any vibration data point within the abnormal rotation cycle as an example, assuming that the first sampling time t1 corresponding to the vibration data point within the abnormal rotation cycle is determined to be the second sampling time from the start of the abnormal rotation cycle, then the first rotation angle of the vibration data point within the abnormal rotation cycle... The following can be calculated: ,in, This is the rotational speed data of the target bearing at the first sampling time. This is the rotational speed data of the target bearing at the second sampling time. The data acquisition interval is defined as follows: within each similar rotation cycle of the abnormal rotation cycle, the first rotation angle is obtained. The corresponding second sampling time t2 is recorded as the vibration data at time t2 within the similar rotation period, which is the matching data of the vibration data of the abnormal rotation period. For the filtering window within the abnormal rotation period, in the similar rotation period of the abnormal rotation period, with the matching data of any vibration data as the center, a number of adjacent matching data of the same size as the filtering window are selected to obtain the matching window corresponding to the similar rotation period.

[0061] In step S170, the filtering weight of each matching window is determined, and the abnormal rotation cycle is filtered through the matching window within the filtering window based on the filtering weight, and the vibration data in the abnormal rotation cycle is updated.

[0062] For each vibration data in an abnormal rotation cycle, the embodiments of this application can determine the matching data corresponding to the vibration data in each similar rotation cycle corresponding to the abnormal rotation cycle through the above process. Furthermore, the embodiments of this application can update the corresponding abnormal rotation cycle vibration data based on the obtained multiple matching data.

[0063] In this embodiment of the application, the above-mentioned filtering weight is an index used to measure the degree of influence of each matching window on the filtering result of vibration data within the abnormal rotation cycle filtering window; the larger the value of the filtering weight, the greater the contribution of the matching data in the matching window to the filtering result in the process of updating the vibration data at the corresponding position of the abnormal rotation cycle.

[0064] For example, the above determination of the filtering weights for each matching window can be achieved as follows: determine the average value of the rotational speed data of the target bearing within the filtering window, denoted as the first rotational speed factor; determine the average value of the rotational speed data of the target bearing within the matching window, denoted as the second rotational speed factor; determine the absolute value of the difference between the first rotational speed factor and the second rotational speed factor, denoted as the first filtering factor; calculate the difference in period length between the abnormal rotational period corresponding to the filtering window and the similar rotational period corresponding to the matching window, denoted as the second filtering factor; determine the filtering weight factor corresponding to the matching window based on the first filtering factor and the second filtering factor, and calculate the sum of the filtering weight factors corresponding to each matching window; for each matching window, determine the filtering weight of the matching window as the ratio of the filtering weight factor corresponding to the matching window to the sum of the filtering weight factors.

[0065] Specifically, taking the matching window corresponding to any similar rotation period as an example, the above-mentioned filtering weight factor can be calculated using the following formula: in, This represents the filter weight factor for the current matching window; The second filtering factor mentioned above, the smaller this value, the more similar the vibration characteristics of the current matching window and the filtering window are, and the greater its weight should be in the filtering process (i.e., the filtering weight factor). The larger the value, the more 1 is added to the denominator to avoid the denominator being 0; The first filtering factor mentioned above, the smaller this value, the closer the rotation data of the current matching window and the filtering window are, and the greater its weight should be in the filtering process (i.e., the filtering weight factor). The larger the value, the more 1 is added to the denominator to avoid the denominator being 0.

[0066] Furthermore, after determining the filter weight factors for each matching window, taking the i-th matching window as an example, the filter weight of the i-th matching window can be determined by the following formula: in, Let be the filtering weights for the i-th matching window; Let be the filtering weight factor for the i-th matching window; This is the sum of the filter weight factors of all matching windows corresponding to the current abnormal rotation cycle.

[0067] After determining the filtering weights of each matching window, for example, the above-mentioned filtering of abnormal rotation cycles through matching windows based on the filtering weights is performed within the filtering window to update the vibration data in the abnormal rotation cycle: within the filtering window, the vibration data in the matching window is weighted based on the filtering weights to obtain the filtering result, and the vibration data in the abnormal rotation cycle is updated with the filtering result.

[0068] Specifically, taking the j-th vibration data point with abnormal rotation period within the filtering window as an example, the filtering result of this j-th vibration data point can be calculated using the following formula: in, The filtering result for the j-th vibration data is used in this embodiment of the application to replace the j-th vibration data during the filtering process; The filtering weights are for the i-th matching window corresponding to the abnormal rotation cycle. Let be the matching data corresponding to the j-th vibration data in the i-th matching window; n is the number of matching windows.

[0069] Furthermore, if there are multiple filtering windows in the current abnormal rotation cycle, after calculating the filtering results under each filtering window using the above method, the filtering results of the vibration data in the abnormal rotation cycle are determined by the following method: calculate the average value of the filtering results corresponding to each filtering window, and update the corresponding vibration data in the abnormal rotation cycle with the average value of the filtering results.

[0070] In step S180, the target bearing is identified as having failed based on the updated vibration data within each rotation cycle.

[0071] After updating the vibration data in each abnormal rotation cycle using the above method (the vibration data in the normal rotation cycle is not processed), the system identifies whether the target bearing has failed based on all rotation cycles, including both normal and abnormal rotation cycles. For example, the failure identification process can be implemented as follows: For each vibration data in each rotation cycle, the average value of other vibration data in the rotation cycle, excluding the current vibration data, is calculated, and the absolute value of the difference between the average value and the current vibration data is calculated and recorded as the failure factor; when the failure factor is greater than a third preset threshold, the first matching data of the current vibration data in the previous rotation cycle and the second matching data in the next rotation cycle are determined; if the failure factors corresponding to the first matching data and the second matching data are both greater than the third preset threshold, the target bearing is determined to have failed, and a failure warning is issued.

[0072] In one specific implementation of this application embodiment, the above-mentioned fault identification process can be implemented as follows: collect vibration data within all rotation cycles, wherein the vibration data of abnormal rotation cycles is the result after filtering and updating, and the vibration data of normal rotation cycles is the original collected data (without filtering wave processing); traverse each vibration data within all rotation cycles and calculate the difference between each vibration data and the mean of all other vibration data within the same rotation cycle (i.e., the above-mentioned fault factor); set a preset threshold for fault judgment (i.e., the above-mentioned third preset threshold) based on the fluctuation range of historical normal vibration data; if the fault factor of a certain vibration data is greater than the third preset threshold, then mark the sampling time corresponding to the vibration data as a suspected fault time; for the vibration data marked as a suspected fault time, find its matching data in two adjacent rotation cycles (with the same determination logic as the above-mentioned determination of matching data), and similarly determine whether the sampling time corresponding to the matching data is a suspected fault time; if the sampling times of the corresponding matching data in adjacent rotation cycles are all marked as suspected fault times, then confirm that the target bearing has failed, issue a warning signal, and notify the staff to carry out maintenance to ensure the safe operation of the wind turbine.

[0073] The foregoing mainly describes the solutions provided by the embodiments of this application from a methodological perspective. To achieve the above functions, it includes corresponding hardware structures and / or software modules for executing each function. Those skilled in the art should readily recognize that, based on the units and algorithm steps of the examples described in conjunction with the embodiments disclosed herein, this application can be implemented in hardware or a combination of hardware and computer software. Whether a function is executed in hardware or by computer software driving hardware depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0074] The above description is merely a preferred embodiment of this application and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of disclosure in this application is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-described concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features with similar functions disclosed in this application.

[0075] Furthermore, while the operations are described in a specific order, this should not be construed as requiring these operations to be performed in the specific order shown or in a sequential order. Multitasking and parallel processing may be advantageous in certain environments. Similarly, while several specific implementation details are included in the above discussion, these should not be construed as limiting the scope of this application. Certain features described in the context of individual embodiments may also be implemented in combination in a single embodiment. Conversely, various features described in the context of a single embodiment may also be implemented individually or in any suitable sub-combination in multiple embodiments.

Claims

1. A method for online fault identification to ensure the safety of wind turbine generators, characterized in that, The method includes: Collect vibration and rotational speed data of the target bearing in the wind turbine unit to be identified; The rotation period of the target bearing is determined based on the data acquisition interval and the rotational speed data at each sampling time. For each rotation cycle, the mean rotation speed of the rotation speed data and the total value of the rotation speed change at each sampling time when the rotation speed data changes are obtained, and the anomaly probability is determined based on the mean rotation speed and the total value of the rotation speed change; The rotation cycle with an abnormal probability greater than a first preset threshold is defined as an abnormal rotation cycle, the other rotation cycles are defined as normal rotation cycles, and similar rotation cycles to the abnormal rotation cycle are determined among the normal rotation cycles. The filter window for the abnormal rotation period is determined based on the abnormal probability and the preset initial window. Based on the rotation angle of the target bearing relative to the start time of the cycle at the sampling time corresponding to each vibration data within the abnormal rotation cycle, the matching data of the vibration data at the corresponding position in the similar rotation cycle is determined, and a matching window is generated; Determine the filtering weight of each matching window, and filter the abnormal rotation cycle within the filtering window through the matching window based on the filtering weight, and update the vibration data in the abnormal rotation cycle; The target bearing is identified as having failed based on the updated vibration data within each rotation cycle.

2. The online fault identification method for ensuring the safety of wind turbine units according to claim 1, characterized in that, The determination of the rotation period of the target bearing based on the rotational speed data at each sampling time and the data acquisition interval includes: The initial rotation angle of the target bearing is set to 0. Every time the data acquisition interval is reached, the product of the rotation speed data at the corresponding sampling time and the data acquisition interval is calculated to obtain the cumulative value of the rotation angle at each sampling time relative to the previous sampling time. The accumulated rotation angle values ​​are sequentially added to the initial rotation angle to obtain the current rotation angle corresponding to each sampling time. When the current rotation angle reaches 360 degrees, the time interval formed by the current sampling time and the sampling time corresponding to the initial rotation angle is determined as the rotation period; The current rotation angle is cleared to zero, and the cleared current rotation angle is recorded as the initial rotation angle. The process of calculating the product of the rotation speed data at the corresponding sampling time and the data acquisition interval at each data acquisition interval is repeated until the cleared current rotation angle is recorded as the initial rotation angle, until the total rotation cycle of the target bearing up to the current time is determined.

3. The online fault identification method for ensuring the safety of wind turbine units according to claim 1, characterized in that, For each rotation cycle, the step of acquiring the average rotation speed of the rotation speed data and the total rotation speed change at each sampling time when the rotation speed data changes, and determining the anomaly probability based on the average rotation speed and the total rotation speed change, includes: The rotational speed data of the target bearing at each sampling time within the rotational cycle is obtained and its average value is calculated to obtain the average rotational speed. The sampling time at which the rotational speed data of the target bearing changes is obtained and recorded as the rotational speed change time; The rotational speed change values ​​at each of the aforementioned rotational speed change moments relative to their previous sampling moment are determined and summed to obtain the total rotational speed change value; The product of the average rotational speed and the total rotational speed change is calculated and normalized to obtain the abnormal probability of the rotational cycle.

4. The online fault identification method for ensuring the safety of wind turbine units according to claim 1, characterized in that, Determining a similar rotation period to the abnormal rotation period within the normal rotation period includes: For each of the abnormal rotation cycles, calculate the difference in cycle length between the abnormal rotation cycle and each of the normal rotation cycles; The normal rotation cycle corresponding to the cycle length difference that is less than the second preset threshold is determined as the similar rotation cycle of the abnormal rotation cycle.

5. The online fault identification method for ensuring the safety of wind turbine units according to claim 1, characterized in that, The step of determining the filtering window for the abnormal rotation period based on the abnormal probability and a preset initial window includes: Set a preset initial window, and calculate the product of the window size of the preset initial window and the anomaly probability, rounding it up, and denote it as the window adjustment factor; If the window adjustment factor is even, the window adjustment factor is determined as the window increment; if the window adjustment factor is odd, the sum of the window adjustment factor plus 1 is determined as the window increment. The filtering window is determined based on the preset initial window and the window increment.

6. The online fault identification method for ensuring the safety of wind turbine units according to claim 1, characterized in that, The method of determining the matching data of the vibration data at the corresponding position in the similar rotation cycle based on the rotation angle of the target bearing relative to the start time of the cycle at the sampling time corresponding to each vibration data within the abnormal rotation cycle, and generating a matching window, includes: For each vibration data point within the abnormal rotation cycle, determine the first sampling time corresponding to the vibration data point within the abnormal rotation cycle; Within the first time interval from the start time of the abnormal rotation cycle to the first sampling time, at each data acquisition interval, the product of the rotational speed data at the corresponding sampling time and the data acquisition interval is calculated and accumulated to obtain the first rotation angle of the vibration data within the abnormal rotation cycle. Within each of the similar rotation cycles, the second sampling time corresponding to the first rotation angle within the similar rotation cycle is determined, and the vibration data corresponding to the second sampling time is recorded as the matching data. The matching window is constructed based on the matching data.

7. The online fault identification method for ensuring the safety of wind turbine units according to claim 6, characterized in that, Determining the filtering weights for each of the matching windows includes: The average value of the rotational speed data of the target bearing within the filtering window is determined and denoted as the first rotational speed factor; The average value of the rotational speed data of the target bearing within the matching window is determined and denoted as the second rotational speed factor; The absolute value of the difference between the first speed factor and the second speed factor is determined and denoted as the first filter factor; Calculate the difference in period length between the abnormal rotation period corresponding to the filtering window and the similar rotation period corresponding to the matching window, and denote it as the second filtering factor; The filter weight factor corresponding to the matching window is determined based on the first filter factor and the second filter factor, and the sum of the filter weight factors corresponding to each matching window is calculated. For each matching window, the ratio of the filter weight factor corresponding to the matching window to the sum of the filter weight factors is determined as the filter weight of the matching window.

8. The online fault identification method for ensuring the safety of wind turbine units according to claim 7, characterized in that, The step of filtering the abnormal rotation cycle based on the filter weights within the filter window using the matching window, and updating the vibration data in the abnormal rotation cycle, includes: Within the filtering window, the vibration data within the matching window are weighted based on the filtering weights to obtain a filtering result, and the vibration data in the abnormal rotation cycle are updated using the filtering result.

9. The online fault identification method for ensuring the safety of wind turbine units according to claim 8, characterized in that, The method further includes: When there are multiple filtering windows in the abnormal rotation cycle, the average value of the filtering results corresponding to each filtering window is calculated, and the vibration data in the abnormal rotation cycle is updated with the average value of the filtering results.

10. The online fault identification method for ensuring the safety of wind turbine units according to claim 1, characterized in that, The step of identifying whether the target bearing has failed based on the updated vibration data within each of the rotation cycles includes: For each vibration data in each rotation cycle, calculate the average value of the other vibration data in the rotation cycle except for the current vibration data, and calculate the absolute value of the difference between the average value and the current vibration data, which is denoted as the fault factor. When the fault factor is greater than the third preset threshold, the first matching data of the current vibration data in the previous rotation cycle and the second matching data in the next rotation cycle are determined. If the fault factors corresponding to the first matching data and the second matching data are both greater than the third preset threshold, it is determined that the target bearing has failed and a fault warning is issued.

Images