A wind turbine fault detection method, device and system

Abstract

The present application relates to the technical field of wind turbine fault detection, and specifically to a wind turbine fault detection method, device and system, the method comprising: collecting winding current data, winding temperature data and active power data during the operation of the wind turbine; calculating the current abnormality index at each moment; and then obtaining the current response index at each moment; obtaining the weight coefficient of other moments in the forward cycle of each moment, and then obtaining the weighted slope at each moment; obtaining the unit fault coefficient at each moment based on the value of the current response index and the weighted slope; obtaining the error threshold range based on the unit fault coefficient; and detecting the fault of the wind turbine based on the error threshold range. The present application improves the accuracy of wind turbine fault monitoring.

Description

Technical Field

[0001] The present application relates to the technical field of wind turbine fault detection, and in particular to a method, device and system for wind turbine fault detection. Background Art

[0002] As one of the key components of wind turbines, the generator has a relatively high probability of failure during actual operation and takes a long time to repair due to its complex internal structure and the influence of complex and changeable environmental factors. Therefore, it is of great significance to the safe operation of wind turbines to conduct real-time evaluation of the operating status of the generator and judge its current health status.

[0003] Traditional wind turbine fault monitoring methods usually rely on the change of a single data indicator for fault warning. This method has problems such as insufficient sensitivity, high false alarm rate, and low diagnostic accuracy when dealing with complex and changeable operating environments and multi-factor coupled faults. Traditional monitoring methods often ignore the inherent connection between parameters and the comprehensive characteristics of data changes, resulting in the inability to capture fault characteristics in a timely and accurate manner when critical faults such as winding short circuits occur, thus affecting the efficiency and accuracy of fault diagnosis. Summary of the invention

[0004] In order to solve the above technical problems, the purpose of this application is to provide a wind turbine fault detection method, device and system. The technical solutions adopted are as follows:

[0005] In a first aspect, an embodiment of the present application provides a method for detecting a fault in a wind turbine generator system, the method comprising the following steps:

[0006] Collect winding current data, winding temperature data and active power data during the operation of wind turbines;

[0007] Based on the proportion of the winding temperature data at each moment to the previous adjacent moment, the proportion of the active power data at the previous adjacent moment to each moment, and the value of the winding current data at each moment, the current abnormality index at each moment is obtained;

[0008] Based on the correlation between the winding temperature data and the winding current data in the neighborhood at each moment, the difference between the correlation between the winding temperature data and the winding current data at the previous adjacent moment, and the current anomaly index, the current response index at each moment is obtained;

[0009] The preset number of consecutive moments before each moment is taken as the forward period of each moment;

[0010] Based on the distance between each moment, the weight coefficients of other moments in the forward cycle of each moment are obtained;

[0011] Based on the difference between each moment and the average of all moments in its forward cycle, and the difference between the current response index at each moment and the average of all moments in its forward cycle, a weighted slope at each moment is obtained;

[0012] The unit failure coefficient at each moment is obtained based on the magnitude of the current response index and the weighted slope;

[0013] An error threshold range is obtained based on the unit failure coefficient; and a wind turbine failure is detected based on the error threshold range.

[0014] Furthermore, the formula of the current anomaly index is:

[0015]

[0016] In the formula, express Current anomaly index at the moment; express The winding current data at the moment, and Respectively Moment and Winding temperature data at the moment, and Respectively Moment and Active power data at the moment.

[0017] Furthermore, the current response index is obtained by:

[0018] For the winding temperature data and winding current data at each moment, starting from the current moment, respectively obtain a sequence of winding temperature data and winding current data at a preset number of consecutive moments in a positive time sequence, and record them as a winding temperature local data sequence and a winding current local data sequence at each moment, respectively;

[0019] The formula for the current response index is:

[0020]

[0021] In the formula, is the current response index at time t; for The correlation coefficient between the winding temperature local data sequence and the winding current local data sequence at the time, for The correlation coefficient between the winding temperature local data sequence and the winding current local data sequence at the time instant; express The current anomaly index at the moment; e is a natural constant.

[0022] Furthermore, the method for obtaining the weight coefficient is:

[0023] The reciprocal of the difference between each moment and the other moments in its forward cycle is used as the weight coefficient of the other moments in the forward cycle of each moment.

[0024] Further, the formula of the weighted slope is:

[0025]

[0026] In the formula, is the weighted slope at time t; is the weight coefficient of the current response index at the i-th moment in the forward cycle at time t, It is expressed as the i-th moment in the forward cycle at time t, represents the mean of all moments in the forward period at time t, Indicates the first The current response index, represents the average value of the current response index in the forward cycle at time t, is the total number of moments in the forward cycle at time t.

[0027] Further, the method for obtaining the unit failure coefficient is:

[0028] For each moment, a calculation result of an exponential function with a natural constant as the base and a weighted slope as the exponent is calculated, and the product of the current response index and the calculation result is used as the unit failure coefficient at each moment.

[0029] Furthermore, the method for obtaining the error threshold range is:

[0030] Calculate the average value and standard deviation of the unit failure coefficient at all times, calculate the multiplication result of the preset constant and the standard deviation, use the sum of the average value and the multiplication result as the upper limit of the error threshold range, and use the difference between the average value and the multiplication result as the lower limit of the error threshold range.

[0031] Further, the fault detection of the wind turbine generator system based on the error threshold range includes:

[0032] For the unit failure coefficient at each moment, when the unit failure coefficient is not within the error threshold range, the wind turbine unit fails; otherwise, the wind turbine unit does not fail.

[0033] In a second aspect, an embodiment of the present application provides a wind turbine fault detection system, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor implements the steps of any one of the above-mentioned wind turbine fault detection methods when executing the computer program.

[0034] In a third aspect, an embodiment of the present application provides a wind turbine fault detection device, the device comprising:

[0035] A data acquisition module is used to collect winding current data, winding temperature data and active power data during the operation of the wind turbine;

[0036] The unit fault coefficient acquisition module is used to obtain the current anomaly index at each moment based on the proportion of the winding temperature data at each moment to the previous adjacent moment, the proportion of the active power data at the previous adjacent moment to each moment, and the value of the winding current data at each moment; obtain the current response index at each moment based on the correlation between the winding temperature data and the winding current data in the neighborhood of each moment, the difference between the correlation between the winding temperature data and the winding current data at the previous adjacent moment, and the current anomaly index; take the preset number of consecutive moments before each moment as the forward cycle of each moment; obtain the weight coefficients of other moments in the forward cycle of each moment based on the distance between each moment; obtain the weighted slope at each moment based on the difference between each moment and the mean of all moments in its forward cycle, and the difference between the current response index at each moment and the mean of the current response index of all moments in its forward cycle; obtain the unit fault coefficient at each moment based on the value of the current response index and the weighted slope;

[0037] The deformation measurement module obtains an error threshold range based on the unit fault coefficient; and detects the fault of the wind turbine unit based on the error threshold range.

[0038] This application has at least the following beneficial effects:

[0039] The present application obtains the winding current data, winding temperature data and generator active power data of the working process of the wind turbine generator through the SCADA system, and calculates the current response index through the relationship between the data. The beneficial effect is that the current response index takes into account the data change characteristics of the winding current data, winding temperature data and generator active power data when a fault occurs, and amplifies the sensitivity of the winding current data fluctuation when the winding is short-circuited; secondly, based on the trend change of the current response index within a cycle when the fault occurs, a unit fault coefficient index is constructed. The unit fault coefficient index fully considers the impact of data at different time points on the slope, and allocates weights according to the distance of the data point from the current moment, so that the data closer to the current moment has a greater impact on the slope. The beneficial effect is that the unit fault coefficient can more accurately reflect the abnormal fluctuation of data at each moment, avoiding the situation where the error of directly calculating the fluctuation of a single data within an operating cycle is large, and improving the accuracy of wind turbine fault monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present application or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0041] Figure 1 A flow chart of the steps of a wind turbine fault detection method provided in one embodiment of the present application;

[0042] Figure 2 A wind turbine data change trend diagram provided for one embodiment of the present application. DETAILED DESCRIPTION

[0043] In order to further explain the technical means and effects adopted by the present application to achieve the predetermined invention purpose, the following is a detailed description of a wind turbine fault detection method, device and system proposed in the present application, its specific implementation method, structure, features and effects, in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0044] Unless defined otherwise, 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 belongs.

[0045] The specific scheme of a wind turbine fault detection method, device and system provided by the present application is described in detail below with reference to the accompanying drawings.

[0046] See also Figure 1 , which shows a flow chart of a wind turbine fault detection method provided by an embodiment of the present application, the method comprising the following steps:

[0047] Step S001, collecting winding current data, winding temperature data and active power data during the operation of the wind turbine generator set.

[0048] The inter-turn short circuit fault of the winding is one of the main fault types of the wind turbine generator, which seriously affects the reliability and safety of the wind turbine operation. Therefore, the present application diagnoses the short circuit fault of the winding of the wind turbine generator.

[0049] This application obtains historical operation data of wind turbines through the data acquisition and monitoring control system (SCADA). The wind turbine operation data collected by the SCADA system is very rich. In order to avoid redundant data interfering with the accuracy of wind turbine generator fault diagnosis, it is necessary to select data directly related to the short-circuit fault of the wind turbine generator winding for analysis.

[0050] When a winding short circuit fault occurs, the short circuit current will increase sharply, thereby generating more heat to increase the winding temperature and reduce the active power of the generator. Therefore, this embodiment selects winding current data, winding temperature data and active power data from the acquired SCADA system data. The sequence lengths of the acquired winding current data, winding temperature data and active power data are all The time interval of the collected data is The specific data sequence length and the time interval for collecting data are determined by the historical operation time and sampling interval of the SCADA system. In this embodiment, , .

[0051] Since the winding current data, the winding temperature data and the active power data have different dimensions, in order to facilitate subsequent analysis, the winding current data, the winding temperature data and the active power data are normalized.

[0052] At this point, the preprocessed winding current data, winding temperature data and active power data are obtained.

[0053] Step S002, based on the proportion of each moment to the winding temperature data of the previous adjacent moment, the proportion of the previous adjacent moment to the active power data of each moment, and the value of the winding current data at each moment, obtain the current abnormality index at each moment.

[0054] When a short circuit occurs in the generator winding of a wind turbine, the short circuited part will form an additional current path, causing the current in the winding to increase. Therefore, when analyzing the short circuit fault of the generator winding of a wind turbine, the abnormal change of the winding current can be detected. However, only using the information of a single variable is not enough to make a comprehensive evaluation of the working condition of the generator winding due to the lack of information. When the winding current of the wind turbine generator increases, the generator winding temperature and the generator active power collected by the SCADA system will also change accordingly. Among them, the wind turbine data change trend diagram is shown in the figure below. Figure 2 shown. Figure 2 The straight line in the middle represents the trend of winding current changing with time, the dotted line represents the trend of winding temperature changing with time, and the dotted line represents the trend of active power changing with time. When a short circuit fault occurs in the generator winding, the winding current and winding temperature show an increasing trend, and the active power shows a decreasing trend.

[0055] Since the generator winding is short-circuited, the current in the winding increases, so the resistance in the winding will generate more heat and thus increase the winding temperature. And when the wind turbine generator winding is short-circuited, the electromagnetic relationship inside the generator changes, the output power of the generator will be affected, and the short-circuit current will consume a lot of active power, which will cause the active power output of the generator to the grid to drop significantly. Therefore, further analysis can be performed based on the correlation between the data.

[0056] When the generator winding of a wind turbine is short-circuited, the winding current increases, while the winding temperature and active power increase and decrease respectively. Based on this, the current abnormality index of the generator winding of a wind turbine at time t is constructed. , the formula is:

[0057]

[0058] In the formula, express Current anomaly index at the moment; express The winding current data at the moment, and Respectively Moment and Winding temperature data at the moment, and Respectively Moment and Active power data at the moment.

[0059] It should be noted that when the wind turbine generator winding is working normally, the collected winding current, winding temperature and active power data are relatively stable, so the calculated current abnormality index It is also relatively stable; when When the generator winding of the wind turbine is short-circuited, The winding temperature data collected at all times increases, thereby increasing the calculated increases; this results in a decrease in the active power data collected at time t, which in turn makes the calculated Increase; thus making the calculated Further increase; and when the wind turbine winding short circuit occurs, The winding current collected at all times will also increase, which will make the calculated current abnormality index Increases, amplifying the sensitivity of winding current data fluctuations when the winding is short-circuited.

[0060] Step S003, based on the correlation between the winding temperature data and the winding current data in the neighborhood of each moment, the difference between the correlation between the winding temperature data and the winding current data at the previous adjacent moment, and the current anomaly index, the current response index at each moment is obtained.

[0061] Furthermore, under normal circumstances, there is a stable relationship between the temperature change and current change of the wind turbine generator winding. However, when the wind turbine generator winding is short-circuited, the collected winding current will increase rapidly, but the winding temperature data has a time lag effect, that is, when the wind turbine generator winding is short-circuited, the winding temperature will respond after a period of time.

[0062] According to the above analysis, for the winding temperature data and winding current data at each moment, from the current moment onwards, obtain the continuous The winding temperature data and winding current data at each moment are sequenced in time order, and are recorded as the winding temperature local data sequence at each moment. And the winding current local data sequence In this embodiment The value of is 60, and the implementer can select other values ​​according to the actual situation. In particular, when the number of consecutive moments obtained forward is less than , pad the sequence with zeros until the number of elements in the sequence is .

[0063] Furthermore, the local data sequence of the winding temperature at each moment is calculated and winding current local data sequence The correlation coefficient between , through the adjacent time correlation coefficient The difference between them reflects the temperature delay response degree. This embodiment uses the Pearson correlation coefficient as the measure of the correlation coefficient, and the implementer can select other correlation coefficients according to actual conditions. Among them, the Pearson correlation coefficient belongs to the well-known technology and will not be described in detail.

[0064] Therefore, based on the above analysis, combined with the current anomaly index , construct the current response index at each moment, the formula is:

[0065]

[0066] In the formula, is the current response index at time t; for The correlation coefficient between the winding temperature local data sequence and the winding current local data sequence at the time, for The correlation coefficient between the winding temperature local data sequence and the winding current local data sequence at the time instant; express The current anomaly index at the moment; e is a natural constant.

[0067] It should be noted that when the wind turbine generator winding is operating normally, the local series data of the winding temperature at each moment And the winding current local sequence data However, when a short circuit occurs in the generator winding of a wind turbine, the winding current increases rapidly, while the rise in winding temperature lags behind the change in winding current, which leads to an increase in the difference in the correlation between the winding temperature and current local sequence data at adjacent moments, that is, Increase; thus making the calculated current response index Increases. This indicates that the possibility of current interruption in the generator winding of the wind turbine is greater at this time.

[0068] Step S004, taking a preset number of consecutive moments before each moment as the forward cycle of each moment; obtaining the weight coefficients of other moments in the forward cycle of each moment based on the distance between each moment; obtaining the weighted slope of each moment based on the difference between each moment and the mean of all moments in its forward cycle, and the difference between the current response index of each moment and the mean of all moments in its forward cycle; obtaining the unit failure coefficient of each moment based on the value of the current response index and the weighted slope.

[0069] Furthermore, taking time t as the end point, obtain the forward and backward The consecutive moments are taken as the forward cycle at time t.

[0070] Specifically, the variation trend of the current response index at all times in the forward cycle at time t is analyzed. Under normal circumstances, the variation trends of the winding current data, the winding temperature data, and the winding power data in the forward cycle are relatively stable.

[0071] Furthermore, in order to consider the changing trend of the data at each moment in the forward cycle, a weighted method is used for processing. Specifically, in this application, weights are assigned according to the time distance between the data point and the moment t, so that the data points closer to the current moment have a greater impact on the changing trend. According to the above analysis, first, the weight coefficient of each moment is constructed, and the formula is: ; In the formula, is the weight coefficient of the i-th moment in the forward cycle at time t; t represents time t, is the i-th moment in the forward cycle at time t.

[0072] It should be noted that when the time when the current response index is located is closer to the current time, the influence on the change trend of the data at the current time is greater, and the weight coefficient obtained at this time is larger; otherwise, the weight coefficient is smaller.

[0073] Furthermore, according to the weight coefficients allocated to each moment, the weighted slope of each moment is constructed, and the formula is:

[0074]

[0075] In the formula, is the weighted slope at time t; is the weight coefficient of the current response index at the i-th moment in the forward cycle at time t, It is expressed as the i-th moment in the forward cycle at time t, represents the mean of all moments in the forward period at time t, Indicates the first The current response index, represents the average value of the current response index in the forward cycle at time t, is the total number of moments in the forward cycle at time t.

[0076] Finally, according to the weighted slope and current response index at each moment, the unit failure coefficient at each moment is obtained, and the formula is:

[0077]

[0078] In the formula, is the unit failure coefficient at time t; is the current response index at time t, e is a natural constant, is the weighted slope at time t.

[0079] It should be noted that when When a short circuit occurs in the generator winding of the wind turbine, The current response index calculated at the moment will show an upward trend, so that The current response index calculated one cycle before the moment The weighted slope Increase, thereby making Unit failure coefficient calculated at all times Increase.

[0080] At this point, the unit failure coefficient at each moment is obtained.

[0081] Step S005, obtaining an error threshold range based on the unit fault coefficient; and detecting the fault of the wind turbine unit based on the error threshold range.

[0082] The set of unit failure coefficients at all times is taken as the unit failure coefficient set.

[0083] When a short-circuit fault occurs in the generator winding of the wind turbine, the unit fault coefficient set further amplifies the fluctuation of the winding current through the relationship between the winding current data, winding temperature data and active power data. According to the unit fault coefficient, the current working condition of the generator winding of the wind turbine can be more accurately reflected.

[0084] Furthermore, in order to analyze and identify abnormal patterns in the data, the computer group failure coefficient set In data analysis, if a data point exceeds three standard deviations of the mean, then this data point is likely to be an outlier caused by an error or an abnormal event. Therefore, the formula for setting the error threshold range in this embodiment is:

[0085]

[0086] In the formula, is the error threshold; Represents the set of computer group failure coefficients The average value of Represents the set of computer group failure coefficients The standard deviation of is the upper limit of the error threshold range, is the lower limit of the error threshold range.

[0087] Under normal operating conditions, the winding current, winding temperature and active power data collected by the wind turbine SCADA system maintain a stable correlation. This stability is reflected in the unit failure coefficient obtained through calculation. It usually shows a relatively stable fluctuation, and the unit failure coefficient However, when a short circuit occurs in the generator winding of a wind turbine, the correlation between these variables will be significantly broken, making Unit failure coefficient calculated at all times Three standard deviations beyond the mean. The unit failure coefficient exceeds the error threshold range When the working state of the current wind turbine generator winding has deviated from the normal mode, a warning is issued; otherwise, the working state of the wind turbine generator winding is normal.

[0088] Based on the same inventive concept as the above method, an embodiment of the present application also provides a wind turbine fault detection system, including a memory, a processor, and a computer program stored in the memory and running on the processor, and when the processor executes the computer program, the steps of any one of the above-mentioned wind turbine fault detection methods are implemented.

[0089] Similarly, based on the same inventive concept as the above method, an embodiment of the present application provides a wind turbine fault detection device, the device comprising:

[0090] A data acquisition module is used to collect winding current data, winding temperature data and active power data during the operation of the wind turbine;

[0091] The unit fault coefficient acquisition module is used to obtain the current anomaly index at each moment based on the proportion of the winding temperature data at each moment to the previous adjacent moment, the proportion of the active power data at the previous adjacent moment to each moment, and the value of the winding current data at each moment; obtain the current response index at each moment based on the correlation between the winding temperature data and the winding current data in the neighborhood of each moment, the difference between the correlation between the winding temperature data and the winding current data at the previous adjacent moment, and the current anomaly index; take the preset number of consecutive moments before each moment as the forward cycle of each moment; obtain the weight coefficients of other moments in the forward cycle of each moment based on the distance between each moment; obtain the weighted slope at each moment based on the difference between each moment and the mean of all moments in its forward cycle, and the difference between the current response index at each moment and the mean of the current response index of all moments in its forward cycle; obtain the unit fault coefficient at each moment based on the value of the current response index and the weighted slope;

[0092] The deformation measurement module obtains an error threshold range based on the unit fault coefficient; and detects the fault of the wind turbine unit based on the error threshold range.

[0093] It should be noted that the above sequence of the embodiments of the present application is for description only and does not represent the advantages and disadvantages of the embodiments. The above is a description of a specific embodiment of this specification. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0094] The various embodiments in the present application are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

[0095] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present application should be included in the protection scope of the present application.

Claims

1. A wind turbine fault detection method, characterized in that: The method comprises the following steps: Collect winding current data, winding temperature data and active power data during the operation of wind turbines; Based on the proportion of the winding temperature data at each moment to the previous adjacent moment, the proportion of the active power data at the previous adjacent moment to each moment, and the value of the winding current data at each moment, the current abnormality index at each moment is obtained; Based on the correlation between the winding temperature data and the winding current data in the neighborhood at each moment, the difference between the correlation between the winding temperature data and the winding current data at the previous adjacent moment, and the current anomaly index, the current response index at each moment is obtained; The preset number of consecutive moments before each moment is used as the forward period of each moment; based on the distance between each moment, the weight coefficients of other moments in the forward period of each moment are obtained; Based on the difference between each moment and the average of all moments in its forward cycle, and the difference between the current response index at each moment and the average of all moments in its forward cycle, a weighted slope at each moment is obtained; Obtaining the unit fault coefficient at each moment based on the magnitude of the current response index and the weighted slope value; obtaining the error threshold range based on the unit fault coefficient; and detecting the fault of the wind turbine unit based on the error threshold range; The formula of the current anomaly index is: In the formula, express Current anomaly index at the moment; express The winding current data at the moment, and Respectively Moment and Winding temperature data at the moment, and Respectively Moment and Active power data at the moment; The method for obtaining the current response index is: For the winding temperature data and winding current data at each moment, starting from the current moment, respectively obtain a sequence of winding temperature data and winding current data at a preset number of consecutive moments in a positive time sequence, and record them as a winding temperature local data sequence and a winding current local data sequence at each moment, respectively; The formula for the current response index is: In the formula, is the current response index at time t; for The correlation coefficient between the winding temperature local data sequence and the winding current local data sequence at the time, for The correlation coefficient between the winding temperature local data sequence and the winding current local data sequence at the time instant; express The current anomaly index at the moment; e is a natural constant.

2. A wind turbine fault detection method according to claim 1, characterized in that: The method for obtaining the weight coefficient is: The reciprocal of the difference between each moment and the other moments in its forward cycle is used as the weight coefficient of the other moments in the forward cycle of each moment.

3. A wind turbine fault detection method according to claim 1, characterized in that: The formula for the weighted slope is: In the formula, is the weighted slope at time t; is the weight coefficient of the current response index at the i-th moment in the forward cycle at time t, It is expressed as the i-th moment in the forward cycle at time t, represents the mean of all moments in the forward period at time t, Indicates the first The current response index, represents the average value of the current response index in the forward cycle at time t, is the total number of moments in the forward cycle at time t.

4. A wind turbine fault detection method according to claim 1, characterized in that: The method for obtaining the unit failure coefficient is: For each moment, a calculation result of an exponential function with a natural constant as the base and a weighted slope as the exponent is calculated, and the product of the current response index and the calculation result is used as the unit failure coefficient at each moment.

5. A wind turbine fault detection method according to claim 1, characterized in that: The method for obtaining the error threshold range is: Calculate the average value and standard deviation of the unit failure coefficient at all times, calculate the multiplication result of the preset constant and the standard deviation, use the sum of the average value and the multiplication result as the upper limit of the error threshold range, and use the difference between the average value and the multiplication result as the lower limit of the error threshold range.

6. A wind turbine fault detection method according to claim 1, characterized in that: The detecting of the fault of the wind turbine generator set based on the error threshold range includes: For the unit failure coefficient at each moment, when the unit failure coefficient is not within the error threshold range, the wind turbine unit fails; otherwise, the wind turbine unit does not fail.

7. A wind turbine fault detection system, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, the steps of a wind turbine fault detection method as described in any one of claims 1 to 6 are implemented.

8. A wind turbine fault detection device, implementing a wind turbine fault detection method as claimed in claim 1, characterized in that: The detection device comprises: A data acquisition module is used to collect winding current data, winding temperature data and active power data during the operation of the wind turbine; The unit fault coefficient acquisition module is used to obtain the current anomaly index at each moment based on the proportion of the winding temperature data at each moment to the previous adjacent moment, the proportion of the active power data at the previous adjacent moment to each moment, and the value of the winding current data at each moment; obtain the current response index at each moment based on the correlation between the winding temperature data and the winding current data in the neighborhood of each moment, the difference between the correlation between the winding temperature data and the winding current data at the previous adjacent moment, and the current anomaly index; take the preset number of consecutive moments before each moment as the forward cycle of each moment; obtain the weight coefficients of other moments in the forward cycle of each moment based on the distance between each moment; obtain the weighted slope at each moment based on the difference between each moment and the mean of all moments in its forward cycle, and the difference between the current response index at each moment and the mean of the current response index of all moments in its forward cycle; obtain the unit fault coefficient at each moment based on the value of the current response index and the weighted slope; The deformation measurement module obtains an error threshold range based on the unit fault coefficient; and detects the fault of the wind turbine unit based on the error threshold range.

Images