Early warning method and device for deterioration of components of wind turbine generator

Abstract

The invention provides a wind turbine generator component degradation early warning method and device, and the method comprises the steps: analyzing the thermal characteristics of a plurality of key components in a fan in a heat balance process, and determining a key parameter abnormality recognition mechanism; in combination with the influence of the wind speed and the environment temperature on the key parameters and the dynamic characteristics of the key parameters, the change characteristics of the key parameters are analyzed; the incidence relation between the wind speed change rate and the key parameter change characteristics is constructed, the key parameter change characteristics, the incidence relation and change characteristic analysis data serve as training data, and a degraded part anomaly recognition model is trained; and acquiring an actual value of the component and a predicted value output by the model, calculating a residual index and an abnormal point rate according to the actual value and the predicted value to perform abnormality judgment, and executing degradation early warning processing when the component is judged to be abnormal. According to the method, data mining and thermal characteristic analysis are combined to construct a degraded part identification model, and a degraded trend part can be positioned timely and accurately.

Description

Technical Field

[0001] The present application relates to the technical field of wind turbine operation and maintenance, and particularly to a method and device for warning of deterioration of wind turbine components. Background Art

[0002] With the continuous increase in the installed capacity of wind power, at the present stage, wind power has been developing towards large-scale, digital and intelligent. At the same time, for wind farms, improving quality and efficiency has become the primary goal of wind farm operation. An important way to achieve this goal is to improve the operation and maintenance quality of the site and the ability of fault warning and diagnosis, so as to achieve the effect of cost reduction in the wind farm.

[0003] In the related art, when performing operation and maintenance on wind turbines, generally, a Supervisory Control And Data Acquisition (SCADA) system is used for fault alarm. After a fault occurs in the wind turbine, on-site operation and maintenance personnel can control the wind turbine to stop, so as to carry out maintenance work and avoid greater losses caused by the aggravation of the fault.

[0004] However, the operation and maintenance method in the above related art can only determine the fault after the fault occurs, and then through the investigation and analysis of the operation personnel, find out the cause of the fault, so as to eliminate the alarm and complete the maintenance work. This operation and maintenance mode avoids the occurrence of greater and more serious faults, but it is unable to actively adjust the operation strategy before the fault occurs to avoid the fault.

[0005] Therefore, how to give a warning against the existing deterioration trend before the components of the wind turbine fail has become an urgent problem to be solved at present. Summary of the Invention

[0006] The present application aims to solve at least one of the technical problems in the related art to some extent.

[0007] To this end, the first object of the present application is to propose a method for warning of deterioration of wind turbine components. The method constructs a deterioration component recognition model by combining data mining and thermal characteristic analysis, and can accurately locate the components with deterioration trends before the fault occurs.

[0008] The second object of the present application is to propose a device for warning of deterioration of wind turbine components;

[0009] The third object of the present application is to propose an electronic device;

[0010] The fourth object of the present application is to propose a non-transitory computer-readable storage medium.

[0011] To achieve the above object, the first aspect of the present application is to propose a method for warning of deterioration of wind turbine components, the method including the following steps:

[0012] Analyze the thermal characteristics of multiple key components in a wind turbine during the thermal balance process, determine multiple key parameters related to the wind speed, and an abnormal identification mechanism based on the multiple key parameters;

[0013] According to the abnormal identification mechanism, analyze the variation characteristics of the multiple key parameters by combining the influence of the wind speed and ambient temperature on the key parameters and the dynamic characteristics of the key parameters;

[0014] Construct the correlation relationship between the wind speed change rate and the variation characteristics of the multiple key parameters, and construct an abnormal identification model for deteriorated components. Use the variation characteristics of the multiple key parameters, the correlation relationship, and the variation characteristic analysis data as training data to train the abnormal identification model for deteriorated components;

[0015] Obtain the actual value of the component to be detected and the predicted value output by the trained abnormal identification model for deteriorated components, calculate the residual index and abnormal point rate based on the actual value and the predicted value. When it is determined that the component to be detected is abnormal according to the residual index and the abnormal point rate, perform a deterioration warning process on the component to be detected.

[0016] Optionally, in an embodiment of the present application, after performing the deterioration warning process on the component to be detected, it further includes: analyzing the abnormal parameters of the component to be detected through the Fault Tree Analysis (FTA) algorithm to determine the abnormal cause; analyzing the abnormal cause and the abnormal parameters of the component to be detected through the Failure Mode and Effects Analysis (FMEA) algorithm to obtain an abnormal diagnosis knowledge table, where the abnormal diagnosis knowledge table includes the abnormal events that occur to the component to be detected; performing operation and maintenance on the component to be detected according to the abnormal diagnosis knowledge table.

[0017] Optionally, in an embodiment of the present application, the multiple key parameters include active power and multiple temperature parameters, and the multiple temperature parameters include the temperatures of different units of each key component. Analyzing the variation characteristics of the multiple key parameters includes: determining the Pearson correlation between the wind speed and the temperature parameters according to the temperature values of different units of each key component under different wind speeds; analyzing the proportion of different temperature values in each wind speed interval and the variation law of different temperature values with the wind speed according to the distribution of the values of the multiple temperature parameters in different wind speed intervals; analyzing the proportion of different temperature values in each ambient temperature interval and the variation law of different temperature values with the ambient temperature according to the distribution of the values of the multiple temperature parameters in different ambient temperature intervals; based on the dynamic fluctuation characteristics of the temperature parameters, set the time interval of parameter change, construct a box plot of the temperature parameter change according to the time interval, and analyze the variation trend of the multiple temperature parameters in a long period through the box plot.

[0018] Optionally, in an embodiment of the present application, analyzing the variation characteristics of the multiple key parameters further includes: determining the maximum and minimum values of the temperature parameter within a preset sliding window according to the distribution of the temperature parameter; obtaining multiple extreme points of the temperature parameter within the preset sliding window by calculating local extreme points; and calculating the change rate of the temperature parameter within the preset sliding window by using the maximum value, the minimum value, and the multiple extreme points.

[0019] Optionally, in an embodiment of the present application, obtaining multiple extreme points of the temperature parameter within the preset sliding window by calculating local extreme points includes: smoothing the temperature parameter and the wind speed parameter within the preset sliding window by using the Savitzky-Golay data smoothing algorithm; analyzing the number of maximum values under different smoothing parameters, and respectively determining the smoothing index and the width of the local extreme points corresponding to the temperature parameter and the wind speed parameter; processing the temperature parameter and the wind speed parameter respectively according to the corresponding smoothing index and the width of the local extreme points, and analyzing the frequency distribution of the distances between the obtained extreme points.

[0020] Optionally, in an embodiment of the present application, constructing the abnormal component deterioration recognition model includes: constructing a corresponding abnormal recognition model for each of the key parameters for each of the key components; wherein, constructing a corresponding abnormal recognition model for each of the key parameters includes: respectively constructing a temperature rise abnormal recognition model and a temperature drop abnormal recognition model for each of the key parameters according to two situations where the temperature rate is positive and the temperature rate is negative.

[0021] Optionally, in an embodiment of the present application, calculating the residual index and the abnormal point rate according to the actual value and the predicted value includes: when the actual value is greater than the predicted value, calculating the difference between the actual value and the predicted value, and determining the abnormal threshold according to the residual distribution of the actual value and the predicted value; calculating the residual index according to the difference and the abnormal threshold, where the residual index represents the abnormal degree of the parameter; and calculating the abnormal point rate according to the sliding window of the abnormal component deterioration recognition model and the number of abnormal points within the sliding window.

[0022] To achieve the above object, a second aspect of the present application further provides a warning device for component deterioration of a wind turbine, including the following modules:

[0023] A thermal analysis module, configured to analyze the thermal characteristics of multiple key components in a wind turbine during the thermal balance process, determine multiple key parameters related to the wind speed, and an abnormal recognition mechanism based on the multiple key parameters;

[0024] A parameter change analysis module, configured to analyze the change characteristics of the multiple key parameters according to the abnormal recognition mechanism, in combination with the influence of wind speed and ambient temperature on the key parameters and the dynamic characteristics of the key parameters;

[0025] A construction module, configured to construct the correlation between the wind speed change rate and the change characteristics of the multiple key parameters, and construct a deteriorated component abnormal recognition model, and use the change characteristics of the multiple key parameters, the correlation, and the change characteristic analysis data as training data to train the deteriorated component abnormal recognition model;

[0026] A detection module, configured to obtain the actual value of the component to be detected and the predicted value output by the trained deteriorated component abnormal recognition model, and calculate the residual index and the abnormal point rate according to the actual value and the predicted value. When it is determined that the component to be detected is abnormal according to the residual index and the abnormal point rate, perform a deterioration warning process on the component to be detected.

[0027] To achieve the above object, a third aspect of the present application further provides an electronic device, including:

[0028] A processor;

[0029] A memory for storing executable instructions of the processor;

[0030] Wherein, the processor is configured to execute the instructions to implement the early warning method for the deterioration of the wind turbine components as described in any one of the first aspects above.

[0031] To achieve the above object, a fourth aspect of the embodiments of the present application further provides a non-temporary computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the early warning method for the deterioration of the wind turbine components as described in any one of the first aspects above is implemented.

[0032] The technical solutions provided by the embodiments of the present application at least bring the following beneficial effects: The present application takes the main systems and key components of a wind turbine as the analysis objects, studies their thermal characteristics, and analyzes the variation laws of relevant characteristic parameters during the thermal balance process. Moreover, under different wind speed conditions, the interval fixed-value method and the data smoothing fitting method are used to determine the change rate of the characteristic parameters fitting the wind speed conditions, and finally, the functional relationship between the wind speed change rate and the change rate of the key characteristic parameters under different wind speed conditions within a fixed time window interval is obtained. The operation parameters of a healthy device in a normal operation state are used to train the model, and a deterioration and abnormality recognition model for key systems and key devices is constructed. Through the fault tree of the deterioration mode and the failure mode and effects analysis method, the parameters that will be abnormal when the deterioration occurs are sorted out, so as to realize the precise positioning and diagnostic analysis of the deteriorated system and deteriorated equipment, and provide support for the efficient operation and maintenance guidance of the wind turbine. Thus, the present application combines data mining and thermal characteristic analysis to construct a deteriorated component recognition model, analyzes the deterioration conditions of the key systems and key devices of the wind turbine, can accurately locate the deteriorated trend components before the occurrence of faults, and improves the accuracy and pertinence of component deterioration recognition. It is beneficial to timely detect the abnormalities of the wind turbine, perform the operation processing of the fan in advance, and is beneficial to ensuring the stability and safety of the operation of the wind turbine.

[0033] Additional aspects and advantages of the present invention will be given in part in the following description, become apparent in part from the following description, or be learned through the practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] The above and / or additional aspects and advantages of the present application will become apparent and be readily understood from the following description of the embodiments in conjunction with the drawings, where:

[0035] Figure 1 is a flowchart of a method for warning of component deterioration of a wind turbine proposed in an embodiment of the present application;

[0036] Figure 2 is a schematic diagram of the distribution of key parameters of a generator under different wind speeds proposed in an embodiment of the present application;

[0037] Figure 3 is a schematic diagram of the distribution of key parameters of a gearbox under different wind speeds proposed in an embodiment of the present application;

[0038] Figure 4 is a percentage stacked bar chart of key parameters under different ambient temperatures proposed in an embodiment of the present application;

[0039] Figure 5 is a box plot of the change of key parameters proposed in an embodiment of the present application;

[0040] Figure 6 A schematic diagram of the frequency distribution of the distance between maximum points proposed in the embodiment of the present application;

[0041] Figure 7 A schematic diagram of the principle of deteriorating warning for wind turbine components proposed in the embodiment of the present application;

[0042] Figure 8 A schematic diagram of the structure of a warning device for wind turbine component deterioration proposed in the embodiment of the present application. Detailed implementation manners

[0043] The embodiments of the present invention will be described in detail below. Examples of the embodiments are shown in the accompanying drawings, where the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to explain the present invention and should not be construed as limiting the present invention.

[0044] It should be noted that during the operation of the wind turbine, the deterioration characteristics of several key subsystems are not obvious during operation, and often only when a failure occurs, there will be obvious manifestations of monitoring parameters. In addition, the abnormal characteristics caused by such deterioration will also be reflected in certain specific wind speed ranges. Based on this, the present application proposes a warning method and device for wind turbine component deterioration, which can locate the deteriorating trend components in a timely and accurate manner.

[0045] The warning method and device for wind turbine component deterioration proposed in the embodiment of the present application will be described in detail below with reference to the accompanying drawings.

[0046] Figure 1 A flowchart of a warning method for wind turbine component deterioration proposed in the embodiment of the present application, as Figure 1 shown, the method includes the following steps:

[0047] Step S101, analyze the thermal characteristics of multiple key components in the wind turbine during the thermal balance process, determine multiple key parameters related to the wind speed, and an abnormal identification mechanism based on the multiple key parameters.

[0048] Specifically, the present application takes multiple key components in the wind turbine as the research object, first analyzes the thermal characteristics of multiple key devices in the wind turbine system, and analyzes and confirms the reasons for the change of characteristic parameters according to the principle of thermal balance. Thereby, the change law of relevant characteristic parameters during the thermal balance process can be determined, that is, the operation parameter change mechanism under thermal performance

[0049] Among them, the key components can be the components that need to be focused on during the operation and maintenance of the wind turbine. The key parameters are the characteristic parameters of the wind turbine that fit the wind speed, and the key parameters are used for deterioration and abnormality analysis under different wind speeds. Taking several key components as examples below, the process of thermal characteristic analysis of this application will be exemplarily described.

[0050] As a first example, the thermal characteristic analysis during the operation of the gearbox. Taking the gearbox as an example, during the operation of the wind turbine, during the operation of the gearbox, due to the continuous friction and heat generation between the gears and bearings during relative operation, the temperature of the components continues to rise. The lubricating oil system is used to dissipate heat and cool down the overall gearbox equipment. Therefore, during the operation of the gearbox, the temperature changes of the gearbox body, lubricating oil temperature and other related components can be observed.

[0051] During specific analysis, the heat generation of the gearbox caused by gear loss, bearing loss, wind resistance loss, and seal loss is analyzed. With the help of the cooling medium to dissipate heat, the temperature of the measuring point rises, and finally the process of heat dissipation through the box surface and the cooling system is analyzed.

[0052] As a second example, the thermal characteristic analysis during the operation of the generator. The generator generates electricity through the electromagnetic effect between the rotor and the stator. The generator body inside the generator generates heat, and the bearings generate heat due to friction. At present, when studying the temperature of the generator, it is mainly considered that the stator and rotor winding parts inside the generator generate heat. The components also achieve the purpose of reducing temperature through various heat exchange methods. During the operation of the unit, the heat generated by the generator is related to the magnitude of the wind speed.

[0053] Furthermore, the key parameter abnormality identification mechanism is determined through thermal characteristic analysis.

[0054] Among them, during the energy conversion process of the wind turbine, the temperature of the components rises due to energy loss. Under normal operating conditions, the power generation performance will fluctuate within a certain range. At the same time, the heat generation and heat dissipation of the components follow the heat transfer principle and maintain a dynamic thermal equilibrium state. Once the equipment operation is abnormal, the heat generation may increase or the heat dissipation efficiency may decrease, resulting in a rapid rise in the temperature of the monitoring point or a slowdown in the cooling process. By monitoring the specific temperature indicators of each key component, it is possible to detect whether there is an abnormal situation.

[0055] For example, when the wind speed increases from 5 m / s to 6 m / s, the heat generation of the components follows a specific pattern; and if the time required for the wind speed change is different, it will lead to different manifestations of the component temperature rise rate, which is mainly related to the heat dissipation characteristics and the length of time. Therefore, when using the SCADA system for fault diagnosis, by analyzing the change rate of the key parameters in the historical data, the abnormal conditions of the unit can be effectively identified.

[0056] Thus, a mechanism for identifying component deterioration anomalies by analyzing the change rate of key parameters is obtained.

[0057] Step S102, according to the anomaly identification mechanism, analyze the change characteristics of multiple key parameters in combination with the influence of wind speed and ambient temperature on the key parameters and the dynamic characteristics of the key parameters.

[0058] Specifically, to implement the mechanism for identifying component deterioration anomalies by analyzing the change rate of key parameters, it is necessary to analyze the change characteristics of multiple key parameters. Analyzing the change characteristics of these key parameters mainly includes: distribution characteristics in different wind speed, temperature intervals, and time intervals, as well as the change rules of characteristic parameters within different time intervals. For example, the change rate of characteristic parameters fitting the wind speed condition can be determined by the interval fixed value method and the data smoothing fitting method.

[0059] It should be noted that the quantification of the key parameter change rate is the basis of the deterioration identification model. In some embodiments of the present application, the change rate of the temperature parameter within a period of time can be determined based on the Savitzky-Golay data smoothing algorithm and the local extreme value method, so as to facilitate subsequent synchronization of the key parameter change rate and the wind speed change rate within a period of time.

[0060] In an embodiment of the present application, the multiple key parameters include active power and multiple temperature parameters, and the multiple temperature parameters include the temperatures of different units of each key component. Analyzing the change characteristics of the multiple key parameters includes the following steps: determining the Pearson correlation between the wind speed and the temperature parameter according to the temperature values of different units of each key component under different wind speeds; analyzing the proportion of different temperature values in each wind speed interval and the change rule of different temperature values with the wind speed according to the distribution of the values of the multiple temperature parameters in different wind speed intervals; analyzing the proportion of different temperature values in each ambient temperature interval and the change rule of different temperature values with the ambient temperature according to the distribution of the values of the multiple temperature parameters in different ambient temperature intervals; based on the dynamic fluctuation characteristics of the temperature parameter, set the time interval of parameter change, construct a box plot of the temperature parameter change according to the time interval, and analyze the change trend of the multiple temperature parameters in the long term through the box plot.

[0061] Specifically, in this embodiment, when analyzing the change characteristics of the key parameters, based on the wind speed and wind direction in the unit historical data, data such as wind speed change, wind speed fluctuation, wind direction change, and wind direction change fluctuation are used to quantify the stochastic characteristics of wind energy on the time scale. Different from the strong correlation between active power and wind speed, the operating parameters of the wind turbine have the following differences from the active power (taking the stator temperature of the generator set as an example):

[0062] First, when the wind speed changes, the parameters are affected differently. When the wind speed is less than the rated wind speed, the relationship between the active power and the wind speed is shown in the following formula:

[0063]

[0064] where ρ is the air density, R is the radius of the wind turbine rotor, C p is the power coefficient, v is the wind speed, and λ is the tip speed ratio.

[0065] However, there is no quantitative formula for the relationship between the temperature parameter and the wind speed. The Pearson correlation between the wind speed and the key parameters is shown in Table 1 below:

[0066] Table 1 Pearson correlation table between wind speed and some key parameters

[0067]

[0068] It should be noted that the Pearson correlation only represents the correlation between two parameters over a period of time and does not mean that the correlation between the two parameters always remains unchanged. If the amount of data is large enough, the Pearson correlation will stabilize to a specific value.

[0069] Second, the change range of the active power is from 0 kW to the rated power, and the change range of the temperature parameter generally lies between the ambient temperature and the threshold set by SCADA. Therefore, the change amplitude of the temperature is smaller than that of the active power.

[0070] Third, in the time dimension, the change of the temperature parameter lags behind and is slower than the change of the active power.

[0071] Fourth, there is a standard power at a single time point to measure the deviation of the active power, but due to the influence of multiple factors on the temperature parameter, there is only a reference interval and a maximum value at a single wind speed and ambient temperature.

[0072] Therefore, when identifying anomalies based on the key parameters of the gearbox and the generator in this application, more attention is paid to the influence of the time interval on the dynamic characteristics of the key parameters.

[0073] Furthermore, analyze the influence of the wind speed on the key parameters. The wind speed is the main reason for the change of the key parameters. In this embodiment, first, analyze Figure 2 the distribution of the key parameters of the generator shown below at different wind speeds.

[0074] Specifically, at Figure 2In part (a), through analysis, it can be obtained that the temperature at the driving end of the generator mainly consists of the range of 30°C to 40°C in the wind speed range, followed by 40°C to 50°C. The range of 20°C to 30°C accounts for a relatively small proportion in different wind speed ranges. As the wind speed increases, the proportion of 30°C to 40°C in the wind speed range shows a downward trend. As the wind speed increases, the proportion of 40°C to 50°C in the wind speed range shows an upward trend. The frequency percentages of 50°C to 60°C are roughly the same in the range of 5 m / s to 9 m / s. The frequency percentage of 60°C to 70°C first rises and then falls as the wind speed increases, and the frequency percentage is the largest at 6 m / s to 7 m / s, followed by 7 m / s to 8 m / s. The range of 70°C to 80°C is concentrated in the range of 6 m / s to 9 m / s.

[0075] In Figure 2 Part (b) shows the temperature distribution of the non-driving end. Through analysis, it can be obtained that in each wind speed range, 37°C to 45°C is the main component, and the frequency percentage first rises and then falls as the wind speed range increases, and the frequency percentage is the largest in the range of 5 m / s to 6 m / s; the proportion of 45°C to 53°C in the wind speed range gradually increases; the frequency percentages of 29°C to 37°C are roughly the same in the wind speed range of 1 m / s to 8 m / s, and the frequency percentage decreases in the range of 8 m / s to 10 m / s. The frequency percentage of 53°C to 61°C is relatively small compared to other ranges. The frequency percentage of 53°C to 61°C first rises and then falls as the wind speed increases, and the frequency percentage is the largest at 9 m / s to 10 m / s.

[0076] In Figure 2 In part (c), through analysis, it can be obtained that the temperature of the generator stator shows obvious upward and downward trends in different temperature ranges as the wind speed increases. The range of 69°C to 89°C appears in the wind speed range of 6 m / s to 7 m / s, and then the frequency percentage is the largest in the wind speed range of 8 to 9 m / s. The range of 89°C to 109°C appears in the wind speed range of 8 m / s to 9 m / s, and the frequency percentage is the largest in the range of 9 m / s to 10 m / s, and the proportion in the wind speed range has been increasing.

[0077] In Figure 2 In part (d), through analysis, it can be obtained that as the wind speed increases, the rotational speed of the generator gradually rises from the low-speed range to the high-speed range. In the range of 6 to 7 m / s and later, the high wind speed range mainly consists of the generator rotational speed of 1480 rpm to 1777 rpm.

[0078] Then, in this embodiment, first, the Figure 3 distribution of the key parameters of the gearbox shown under different wind speeds is analyzed.

[0079] Specifically, the temperature distribution of the front-end bearing of the gearbox is as shown in Figure 3As shown in the first part. Through analysis, it can be obtained that in the range of 33°C to 40°C, since the frequency percentage is lower than that in other temperature ranges and cannot be visually seen from the figure, the frequency in the range of 2 m / s to 4 m / s is higher than that in the range of 4 m / s to 7 m / s; within the wind speed range, the frequency of 47°C to 54°C is higher than that of 40°C to 47°C, and the frequency percentage of 47°C to 54°C is the highest in the range of 3 m / s to 4 m / s and then drops rapidly; the frequency percentage of the range of 54°C to 61°C is relatively low in the range of 2 m / s to 3 m / s, then gradually increases with the increase of wind speed, reaches the maximum in the wind speed range of 5 m / s to 6 m / s, and then gradually decreases and exists but is very small in the range of 8 m / s to 9 m / s; the range of 61°C to 68°C appears in the range of 3 m / s to 4 m / s, the frequency percentage is still very low in the range of 4 m / s to 5 m / s, then increases rapidly, reaches the maximum in the range of 7 m / s to 8 m / s, and then the frequency percentage decreases with the increase of the wind speed range, but it is still the main component in each wind speed range; the range of 68°C to 75°C appears in the range of 7 m / s to 8 m / s, and the proportion within the range is very small. With the increase of wind speed, the frequency percentage is relatively high in the ranges of 8 m / s to 9 m / s and 9 m / s to 10 m / s, and then the frequency percentage gradually decreases, and the proportion of the range is the largest in the range of 10 m / s to 11 m / s.

[0080] The temperature distribution of the rear bearing is as Figure 3 shown in the second part. Through analysis, it can be obtained that the high-temperature range of 70°C to 77°C appears in the wind speed range of 5 m / s to 6 m / s, and then the proportion within the wind speed range keeps increasing. In Figure 3 the third part is the lubricating oil inlet oil pressure of the gearbox. With the increase of wind speed, the proportion of the range of 5 bar to 7 bar gradually increases. In Figure 3 the fourth part, the maximum temperature range of the lubricating oil temperature is 60°C to 66°C, which appears in the wind speed range of 6 m / s to 7 m / s, then gradually increases, reaches the maximum in the wind speed range of 8 m / s to 9 m / s, and finally gradually decreases. The wind speed range of 10 m / s to 13 m / s is composed of the temperature range of 54°C to 66°C.

[0081] Based on the distribution analysis of the key parameters of the gearbox and the generator at different wind speeds in this embodiment, it can be obtained that some parameters have obvious interval replacement phenomena with the increase of temperature, and high-temperature parameters generally appear in the intervals with relatively high wind speeds.

[0082] Furthermore, analyze the influence of the ambient temperature on the key parameters. For example, analyze the Figure 4 shown separate situations. Generally, within each ambient temperature range, it is generally composed of several parameter temperature ranges. Among them, the overall proportion of the ambient temperature of 28°C to 32°C is less than 0.15%.

[0083] Specifically, through analysis, it can be obtained that in Figure 4 In part (a), the proportion of the parameter temperature range of 36°C to 40°C is very low and appears in the environmental temperature range of 0°C to 16°C; as the environmental temperature range increases, the low-temperature range of the parameter gradually disappears, and the proportion of the intermediate temperature range gradually increases; the proportion of 72°C to 76°C is the highest in the environmental temperature range of 16°C to 20°C, reaching 14%; when the environmental temperature is 28°C to 32°C, the lowest parameter temperature range is 44°C to 48°C.

[0084] In Figure 4 In part (b), as the environmental temperature range increases, the low-temperature range of the parameter of 32°C to 44°C gradually disappears, the proportion of the temperature range of 60°C to 64°C first increases and then decreases, and the proportion of the temperature range of 56°C to 60°C first decreases, then increases, and then decreases. In Figure 4 In part (c), in different environmental temperature ranges, the formed lubricating oil temperature ranges are roughly the same except in the environmental temperature range of 28°C to 32°C. The low-temperature range of the lubricating oil of 32°C to 36°C mainly appears in the environmental temperature range of 0°C to 4°C, the low temperature of 36°C to 40°C mainly appears in the environmental temperature range of 0°C to 12°C, and the parameter range of 40°C to 44°C mainly appears in the environmental temperature range of 4°C to 24°C. The high-temperature range of 64°C to 68°C mainly appears in the environmental temperature range of 16°C to 24°C.

[0085] In Figure 4 In part (d), within each environmental temperature range, the parameter ranges are relatively concentrated. For example, in the environmental temperature range of 8°C to 12°C, the sum of the parameter temperature ranges of 32°C to 36°C and 36°C to 40°C accounts for 79% of the entire range; the parameter range of 20°C to 24°C appears in the environmental temperature range of 0°C to 8°C; in the environmental range of 24°C to 28°C, the lowest parameter range is 32°C to 36°C. In Figure 4 In part (e), the temperature at the non-drive end has a corresponding relatively high proportion range in different environmental temperature ranges. For example, in the environmental temperature range of 0°C to 4°C, the proportion of the parameter range of 28°C to 40°C is 96%, and in the environmental temperature range of 4°C to 8°C, the proportion of the parameter range of 32°C to 40°C is 86%. The proportion of 12°C to 20°C in each region is small. As the environmental temperature increases, the main component region gradually changes from 36°C to 40°C to 40 to 44°C except for 28 to 32°C. In Figure 4 In part (f), for each environmental temperature range, there are two obvious characteristics, namely having the same main parameter range and being composed of multiple parameter ranges. The parameter range of 48°C to 60°C is the main component of other temperature ranges except the environmental temperature range of 28°C to 32°C.

[0086] In this embodiment, by analyzing the distribution of key parameters in different environmental intervals, it can be determined that the low interval of the parameter generally only appears in the low environmental temperature interval; the distributions of different parameters in the environmental interval have their own characteristics, which are mainly manifested in the different concentration degrees of the intervals of the parameters within the interval and the different numbers of parameter intervals. Therefore, when constructing an anomaly recognition model in the follow-up, the influence of environmental temperature on the parameter distribution needs to be considered.

[0087] Furthermore, analyze the quantization of the dynamic characteristics of key parameters. When analyzing the volatility of the parameters themselves, since the temperature parameter changes with a small amplitude and relatively slowly, when analyzing the fluctuation characteristics of the temperature parameter, the time interval of the parameter change needs to be considered.

[0088] For example, as Figure 5 shown, when the time interval is 5 minutes, from the box plot of the parameter change, it can be seen that the temperature change range of the generator stator is the largest, which is -2.87°C to 3.15°C. Secondly, the temperature change range of the rear bearing of the high-speed shaft of the gearbox is -2.84°C to 2.95°C, and the temperature change ranges of the gearbox lubricating oil temperature, the driving end temperature and the non-driving end temperature of the generator are -1.00°C to 1.00°C.

[0089] By analyzing the box plot of the parameter change, it can be obtained that when the time interval is 5 minutes, the parameter change is small, and the parameter change is slow within a short time. The purpose of analyzing the parameter change in this embodiment is to analyze the change rate of the parameter within a period of time, that is, the rising rate and the falling rate of the parameter. When analyzing the change rate, the overall rising or falling trend of the parameter within a period of time is quantified, and the short-term small-range up and down fluctuations of the parameter are ignored.

[0090] Furthermore, based on the above analysis examples of multiple key parameters, using the distribution data and the obtained rules, calculate the change rate of each key parameter.

[0091] In an embodiment of the present application, analyzing the change characteristics of multiple key parameters further includes: determining the maximum value and the minimum value of the temperature parameter within a preset sliding window according to the distribution of the temperature parameter; obtaining multiple extreme points of the temperature parameter within the preset sliding window by calculating the local extreme points; using the maximum value, the minimum value and the multiple extreme points to calculate the change rate of the temperature parameter within the preset sliding window.

[0092] Specifically, within the preset sliding window H, using the data distribution situation and the obtained data change rules in the above analysis, the maximum value and the minimum value of the parameter within this window can be obtained, and then the change rate can be calculated through the following formula:

[0093]

[0094] Among them, x MAX and x MIN are respectively the maximum and minimum values of the parameter during this period, and t max and t min are respectively the time points corresponding to the maximum and minimum values. However, calculating through the above formula has the defect of reducing the change rate, that is, if there is a maximum value x i , x i < x MAX , then there is the relationship shown in the following formula:

[0095]

[0096] Among them, the reason for this situation is that the maximum value point is close to the minimum value point, and at the same time, the maximum value and the maximum value point are far apart.

[0097] To avoid the absolute value of k from becoming smaller caused by the above situation, within a fixed time period, the local extreme value point method is used to obtain the extreme value points. Let the local extreme value point be the width h. There are multiple maximum value points within the sliding window, which avoids the situation where the absolute value of k is overly reduced. Calculating the change rate in such a form may result in multiple rising rates and falling rates within the sliding window. Finally, the maximum rising rate and the minimum falling rate are used as the rising change rate and the falling change rate in the current time period.

[0098] In this embodiment, by calculating the local extreme value points, multiple extreme value points of the temperature parameter within the preset sliding window are obtained, including: smoothing the temperature parameter and the wind speed parameter within the preset sliding window through the Savitzky-Golay data smoothing algorithm; analyzing the number of maximum values under different smoothing parameters, and respectively determining the smoothing index and the width of the local extreme value point corresponding to the temperature parameter and the wind speed parameter; processing the temperature parameter and the wind speed parameter respectively according to the corresponding smoothing index and the width of the local extreme value point, and analyzing the frequency distribution of the distances between the obtained extreme value points.

[0099] Specifically, data smoothing is first performed based on the Savitzky-Golay algorithm. When analyzing monitoring data, in order to improve the efficiency of finding extreme values and reduce the influence of redundant extreme values on the results, it is necessary to perform preprocessing of data smoothing. In order to retain the useful information of the parameter while performing smoothing processing, this embodiment uses the Savitzky-Golay data smoothing algorithm to smooth the parameter data. This algorithm is a method of data smoothing using a high-order polynomial based on the principle of least squares.

[0100] As an example, assume that there are n = m + 1 equally spaced measurement points, that is, x i(i = -m, -m + 1, …, m - a, m), assuming that the data points within the window can be fitted by a polynomial of degree k - 1, that is, the operation is performed through the following formula:

[0101] y i = a0 + a1i + a2i 2 + … + a k-1 i k-1

[0102] From this, n such equations can be obtained, forming a k - element linear equation system. To solve for k fitting parameters, that is, a j (j = 0, 1, 2, …, k - 1), when n > k, the least - squares method is used for solution, and the calculation formula is as follows:

[0103]

[0104] This formula is represented in matrix form as the following formula:

[0105] Y (2m+1)×1 = X (2m+1)×k · A K×1 + E (2m+1)×1

[0106] Among them, the least - squares solution of A is calculated through the following formula:

[0107]

[0108] The model smoothed value of Y is calculated through the following formula:

[0109]

[0110] Among them, B = X · (X T · X -1 ) -1 · X T . B is related to the X matrix, and the X matrix is determined by the data width n and the degree of the polynomial (k - 1). The above - mentioned least - squares solution calculation formula of A is the data smoothing formula, which is the linear relationship between the smoothed data and the original data Y matrix, contains n expressions, and can be used to calculate the smoothed values of each point within the smoothing window.

[0111] Then, the determination of the local change rate is carried out. For the width of the local extreme point h, considering the temperature parameter change characteristics and the sensitivity to the extreme value, in this embodiment, the number of maximum points of the smoothed data is used to evaluate the data smoothing effect.

[0112] For example, for the analysis of the wind speed and key parameter data of a certain month, processed in the above - mentioned manner, the result is the number of maximum points of the stator temperature L1 after smoothing as shown in Table 2 below.

[0113] Table 2 Statistical table of the maximum points of the generator stator temperature L1 under different smoothing parameters

[0114]

[0115] As can be seen from Table 2, as n increases, the number of extreme points gradually decreases. When determining the final number of maximum points, it is necessary to balance the model accuracy and the impact of reducing redundant maximum values. When determining the distance between the final maximum values, it is necessary to reduce adjacent extreme points because of the redundant extreme points caused by the close distance and the loss of the extreme values caused by the smoothing transition. The evaluation of the smoothing parameter is based on the change of extreme points before and after smoothing and the degree of smoothing of small fluctuations. Considering the sensitivity of the change rate to extreme values, the smoothing parameter of k = 2 is used to preprocess the data. The measurement of the local extreme point h is to reduce the influence of redundant extreme points on the change rate.

[0116] Then analyze the number of maximum values under different smoothing parameters to determine the final model parameters. When the smoothing index n = 13 and h = 1, there are 356 maximum points, and the change compared with 317 maximum points when the smoothing index n = 15 and h = 1 has a smaller change amplitude than that of 812 (n = 5) and 619 (n = 7). After determining n = 5, the number of maximum points when h = 5 is close to the number of maximum points when n = 13. Therefore, considering the smoothing effect, when h = 5, the effect of the extreme points of n = 13 is achieved through the form of local extreme points. Therefore, the temperature parameters are preprocessed with n = 5 and h = 5.

[0117] In the example, the following Table 3 shows the statistics of the maximum points of the wind speed under different parameters.

[0118] Table 3 Statistical table of the maximum points of the wind speed under different smoothing parameters

[0119]

[0120] As can be seen from Table 3, compared with the above temperature parameters, the initial number of maximum points of the wind speed 1534 (n = 5, h = 1) is greater than the initial number of maximum points of the temperature parameters 812 (n = 5, h = 1). As n increases, the number of extreme points gradually decreases. When h = 8, the number of extreme points tends to be stable. Therefore, the wind speed parameters are preprocessed with n = 5 and h = 8.

[0121] According to the processing method in the above example, for the preprocessed data of the wind speed and the stator temperature L1 before and after preprocessing and the corresponding local maxima, after smoothing and local extreme values, the number of maximum points of the wind speed decreases from 164 to 30, and the number of maximum points of the stator temperature L1 decreases from 82 to 22.

[0122] Finally, when analyzing the change rate of the unit based on the key parameters, after determining the maximum value points in the entire region in the above manner, the distance between the extreme value points was also analyzed. The result is as Figure 6 shown. In this embodiment, the median 20 is used as the value of the sliding window M of the subsequent constructed anomaly recognition model.

[0123] Step S103: Establish the correlation between the wind speed change rate and the change characteristics of multiple key parameters, and construct an abnormal component recognition model. Use the change characteristics of multiple key parameters, the correlation, and the change characteristic analysis data as training data to train the abnormal component recognition model.

[0124] Specifically, as can be seen from the above analysis, the distribution of key parameters under wind speed and ambient temperature is different. The wind speed change rate is a parameter related to time. If the wind speed change rate is large, that is, the wind speed rises from a low speed to a high speed in a short time, the heat generated by the component will also increase. However, since the heat dissipation area and the heat transfer coefficient are constant, the temperature of the component will rise. Therefore, in the subsequent construction of the anomaly recognition model in this application, the wind speed change rate is used as an input parameter of the model. In this step, first, using the analysis results of the key parameters in the above embodiment, analyze the reasonable correlation between the wind speed change rate and the key parameter change rate, and determine the correlation time under normal conditions, that is, determine the functional relationship between the wind speed change rate and the change rate of the key characteristic parameters under different wind speed conditions within a fixed time window interval.

[0125] Then, based on the rules sorted out in the above embodiment related to the key parameters, including the change characteristics of the key parameters, the correlation, and the relevant data involved in the change characteristic analysis, establish a corresponding abnormal component parameter recognition model.

[0126] Among them, when constructing the anomaly recognition model in this application, the operating parameters of healthy equipment in the normal operating state are used to train the model to construct an abnormal degradation recognition model for key equipment. Based on the deep learning algorithm, construct an abnormal parameter recognition model. Based on the maximum rising rate and the minimum falling rate of the characteristic parameters in the research interval and the corresponding wind speed change rate, construct a neural network anomaly recognition model, and train the model based on normal historical data.

[0127] In an embodiment of this application, constructing an abnormal component recognition model includes: for each key component, constructing a corresponding anomaly recognition model for each key parameter; among them, constructing a corresponding anomaly recognition model for each key parameter includes: respectively constructing a temperature rise anomaly recognition model and a temperature drop anomaly recognition model for each key parameter according to the two situations where the temperature rate is positive and the temperature rate is negative.

[0128] Specifically, in this embodiment, different anomaly recognition models need to be constructed for different key devices. However, for the same key device, taking a generator as an example, the generator system has three temperature parameters: the temperature at the driving end, the temperature at the non-driving end, and the stator temperature, and there are three corresponding temperature change rates. When constructing the recognition model, although the neural network model can construct a multi-input and multi-output model, if three temperature parameters are input simultaneously during the model construction process, the corresponding relationships of the three parameters also need to be constructed in the model, increasing the complexity of the model. If three parallel prediction models are constructed for the generator and then the anomaly rates of each parameter are calculated to identify the abnormal state of the unit, this is not only beneficial for locating the part where the anomaly occurs but also can identify the transmission of the anomaly in the system through the anomaly rates of the parameters, that is, an anomaly in a single parameter may cause anomalies in multiple parameters.

[0129] During the process of constructing the model, since the temperature rise and the temperature fall are two opposite directions, and in the case where the wind speed and the wind speed change direction are known, the change direction of the temperature still has uncertainty. Therefore, models are constructed separately for the positive temperature rate and the negative temperature rate to avoid misjudgment of the model caused by the uncertainty of the direction change.

[0130] Furthermore, when training the constructed model, during the process of constructing the anomaly recognition model based on the historical data of normal units, the input parameters during the model training process include but are not limited to the wind speed, the wind speed change rate, the most recent wind speed maximum point, the ambient temperature, the parameter temperature, the parameter temperature change rate, and the most recent temperature maximum point in the above embodiments, where the temperature change rate is the target parameter. The average error and the mean absolute error can be used to measure the accuracy of the network model during the training process. The specific training process can refer to the training method of the neural network model in the relevant embodiments, and this application does not limit it.

[0131] Step S104, obtain the actual value of the component to be detected and the predicted value output by the trained anomaly recognition model of the deteriorated component, calculate the residual index and the anomaly point rate according to the actual value and the predicted value, and when it is determined that the component to be detected is abnormal based on the residual index and the anomaly point rate, perform the deterioration early warning process on the component to be detected.

[0132] Specifically, during the actual deterioration anomaly detection, the actual value of the component to be detected regarding a certain key parameter is obtained through the acquisition device, and the predicted value output by the deteriorated component anomaly recognition model based on the current and historical actual operation data of the component is obtained. Real-time monitoring of anomaly recognition and health maintenance are performed according to the actual value and the predicted value.

[0133] As a possible implementation, the abnormal limit can be determined based on the residual distribution of the predicted value and the actual value of the deep learning algorithm, and the degree of abnormality of the unit can be calculated according to the degree of deviation of the residual. The abnormality is judged by calculating the abnormal point rate based on the number of abnormal points in the research interval. The principle of the early warning method for the deterioration of the wind turbine components in the embodiments of the present application is as Figure 7 shown.

[0134] Among them, after determining the model structure of the neural network, when performing abnormal detection on a certain key parameter, if the actual value is less than the predicted value, it means that the value of this key parameter is a normal value. If the actual value of this point is greater than the predicted value, it is necessary to judge whether this point is an abnormal point according to the residual situation. In this example, the abnormal limit is determined based on the residual distribution of the predicted value and the actual value. In order to balance the accuracy of the model and reduce the error of the model, the corresponding value of P99 of the residual is used as the abnormal limit.

[0135] In an embodiment of the present application, calculating the residual index and the abnormal point rate according to the actual value and the predicted value includes: when the actual value is greater than the predicted value, calculating the difference between the actual value and the predicted value, and determining the abnormal threshold according to the residual distribution of the actual value and the predicted value; calculating the residual index according to the difference and the abnormal threshold, where the residual index represents the degree of abnormality of the parameter; calculating the abnormal point rate according to the sliding window of the deterioration component abnormal identification model and the number of abnormal points in the sliding window.

[0136] Specifically, in this embodiment, during the operation of the fan, when performing abnormal detection on the key parameters of a certain key component, in order to avoid misjudgment caused by single-point random error, the positive residual index and the negative residual index are used to evaluate the degree of abnormality of the parameter. The calculation process can be specifically represented by the following formula:

[0137] E i = x i - x' i

[0138]

[0139] where x i is the actual value, x' i is the predicted value, P is the corresponding value of the residual P99, y is the positive residual and the negative residual, M is the length of the sliding window, y' is the abnormal point rate, and M'' is the number of abnormal points. Y is the positive residual index and the negative residual index. Y represents the degree of abnormality of the parameter. If Y < 1, it means that the parameter is normal. If Y > 1, it means that the parameter is abnormal. As Y increases, the degree of abnormality of the parameter increases accordingly.

[0140] Thus, by combining the calculated outlier rate and residual index mentioned above, it can be determined whether a key parameter is an abnormal parameter when the actual value of a certain key parameter is greater than the predicted value. When both the outlier rate and the residual index meet the corresponding judgment conditions, such as both being greater than the corresponding thresholds, it is determined that the key parameter is an abnormal parameter, and there is a deterioration abnormality in the key component.

[0141] Furthermore, a warning is issued for the component determined to have deterioration. For example, after determining that a certain component has a deterioration abnormality, an audible and visual alarm is given, and an alarm message is sent to the relevant operation and maintenance personnel through wireless connection, indicating that there is a deterioration trend in the component and it needs to be processed in a timely manner.

[0142] Based on the above embodiments, as a possible operation and maintenance method for deteriorated components, in an embodiment of the present application, after performing the deterioration warning processing on the component to be detected, it further includes: analyzing the abnormal parameters of the component to be detected through the Fault Tree Analysis (FTA) algorithm to determine the abnormal cause; analyzing the abnormal cause and the abnormal parameters of the component to be detected through the Failure Mode and Effects Analysis (FMEA) algorithm to obtain an abnormal diagnosis knowledge table, where the abnormal diagnosis knowledge table includes the abnormal events that occur in the component to be detected; and performing operation and maintenance on the component to be detected according to the abnormal diagnosis knowledge table.

[0143] Specifically, as Figure 7 shown, first, a mechanism analysis of the health maintenance of the wind turbine is carried out based on the fault tree. The mechanism of abnormal events in the operation state of the wind turbine is analyzed through the FTA algorithm. Taking the abnormal events of subsystems or components during operation as the starting point, the causes leading to the abnormality are listed hierarchically based on the causal relationship to form a logical tree diagram. This method can achieve the following purposes: First, identify the basic events that lead to the abnormal operation state, so as to provide the wind farm operators with the basic reasons for reducing the abnormal events, and achieve the purpose of reducing the abnormal rate. Second, comprehensively and systematically describe various factors and causal relationships of the abnormal operation state. Third, help find out various inherent or potential factors in the wind turbine that are prone to cause abnormalities, and provide a basis for the design, construction, and operation of the wind turbine. Fourth, facilitate the qualitative and systematic evaluation of the abnormal situation of subsystems.

[0144] FTA is based on the conditions and consequences of abnormalities occurring during the operation of the wind turbine, and forms a tree diagram in a logical relationship manner to represent the causal relationship between the elements leading to the abnormal state, and to clarify the possible ways for the occurrence of the abnormal operation state. The steps of performing FTA analysis on the abnormal events in the operation state of the wind turbine in this embodiment are as follows:

[0145] First step: Familiarize with the operation and daily maintenance process of the wind turbine. Analyze in detail the operating status and various operating parameters of the wind turbine. Second step: Collect and determine the abnormal events in the operation status of the wind turbine. Collect abnormal cases, conduct abnormal statistics, and analyze the possible abnormalities that may occur during the operation of the wind turbine. Third step: Determine the top event. The abnormal event in the operation status of the wind turbine that needs to be analyzed is the top event. Fourth step: Determine the target value. Based on expert experience, existing research, and abnormal cases in the operation process of the wind farm, determine the probability (frequency) of the occurrence of abnormal operation status. Fifth step: Investigate the cause events. Investigate all cause events and various factors related to the abnormality. Sixth step: Draw the abnormal tree of the abnormal event in the operation status. Starting from the top event, find out the cause elements layer by layer, and finally form an abnormal tree according to their causal relationship. Seventh step: Simplify the abnormal tree of the operation status and determine the importance of each relevant element. Eighth step: Determine the probability of the occurrence of the abnormal event in the operation status. Determine the occurrence probability of all causes based on existing research, expert experience, or data analysis, and finally obtain the occurrence probability of the top abnormal event. Ninth step: Discuss in terms of adjustable events and non-adjustable events. Determine how to adjust the adjustable events, and determine the occurrence probability for the non-adjustable events.

[0146] Furthermore, carry out health maintenance measures for wind turbines based on FMEA. Since the above FTA is a qualitative or quantitative abnormal analysis method in a logical reasoning manner, the analysis of abnormal events in the operation status of wind turbines based on the FTA method can systematically find out the causes of various abnormalities, and based on probability or expert experience, the impact degree on the whole machine can be analyzed. And when an abnormality occurs, it is necessary to analyze the symptoms of the abnormality, the impact on adjacent components, and the corresponding handling methods, etc. For this reason, this embodiment adopts the Failure Mode And Effects Analysis (FMEA) algorithm to clarify its impact on the system by analyzing the possible abnormal modes of each component. The FMEA method can analyze all possible simple abnormal modes of each component and their impacts on the system and adjacent systems. Since FMEA studies each abnormal mode of each component, the FMEA method is used to conduct qualitative analysis on the health maintenance of components.

[0147] FMEA can evaluate and analyze various abnormal events, which is convenient for wind power researchers and wind farm operators to rely on existing technologies to deal with abnormal events of wind turbines. Apply FMEA to evaluate and analyze the abnormal modes of the operation status of wind turbines, and formulate improvement measures for abnormal modes with greater loss degrees. By applying FMEA to analyze and evaluate abnormal events in the operation status of wind turbines, risks existing in components can be discovered in time, predictive maintenance can be implemented, potential abnormal events of components can be effectively avoided, the operation and maintenance costs of wind turbines can be reduced, and the performance of wind turbines can be improved.

[0148] When applying the FMEA method for operation and maintenance in this embodiment, the following basic implementation steps are included: First step, determine the functions and structure of the wind turbine, determine the reliability requirements of the wind turbine, and master the composition and combination methods of the subsystems of the wind turbine. Second step, determine the scope of the functions and devices of the wind turbine, and pre-determine the analysis level based on operation experience and statistical data of abnormal events. Third step, draw the structure block diagram of the wind turbine, and classify each system of the wind turbine by function, such as the wind energy capture function, transmission function, control function, etc. Draw structure blocks with the above functions as units, and consider the relationships (series and parallel) of various functions between the blocks. Fourth step, obtain the abnormal operation mode corresponding to each block of the wind turbine. Fifth step, select the abnormal mode for analysis. Sixth step, list the relevant causes of the abnormal events in the operation state of the wind turbine. Determine the treatment measures according to expert experience, daily operation reports, component non-conformance reports, and previous FTA analysis results.

[0149] Thus, in this embodiment, the abnormal components detected in a timely manner are maintained, and the faults can be eliminated in a timely manner.

[0150] In summary, the early warning method for the deterioration of the components of the wind turbine in the embodiment of the present application takes the main systems and key components of the wind turbine as the analysis objects, studies their thermal characteristics, and analyzes the variation laws of the relevant characteristic parameters in the thermal balance process. And, under different wind speed conditions, the interval fixed value method and the data smoothing fitting method are used to determine the change rate of the characteristic parameters fitting the wind speed conditions, and finally the functional relationship between the wind speed change rate and the change rate of the key characteristic parameters under different wind speed conditions within a fixed time window interval. Train the model with the operation parameters of healthy equipment in the normal operation state, construct the deterioration abnormal recognition model of the key system and key equipment, and through the fault tree and fault failure mode and effect analysis method of the deterioration mode, sort out the parameters that will be abnormal when the deterioration occurs, so as to realize the accurate positioning and diagnostic analysis of the deteriorated system and deteriorated equipment, and provide support for the efficient operation and maintenance guidance of the wind turbine. Thus, this method combines data mining and thermal characteristic analysis to construct a deteriorated component recognition model, analyzes the deterioration of the key systems and key equipment of the wind turbine, and can accurately locate the deteriorated trend components before the occurrence of faults, improving the accuracy and pertinence of component deterioration recognition. It is beneficial to timely detect the abnormalities of the wind turbine and perform the operation processing of the fan in advance, which is beneficial to ensuring the stability and safety of the operation of the wind turbine.

[0151] To implement the above embodiment, the present application also proposes an early warning device for the deterioration of the components of the wind turbine. Figure 8 Shown in the following is the structural schematic diagram of an early warning device for the deterioration of the components of the wind turbine proposed in the embodiment of the present application. Figure 8As shown in the figure, the device includes a thermal analysis module 100, a parameter change analysis module 200, a construction module 300, and a detection module 400.

[0152] Among them, the thermal analysis module 100 is used to analyze the thermal characteristics of multiple key components in the wind turbine during the thermal balance process, determine multiple key parameters related to the wind speed, and an abnormal identification mechanism based on the multiple key parameters.

[0153] The parameter change analysis module 200 is used to analyze the change characteristics of multiple key parameters according to the abnormal identification mechanism, combining the influence of the wind speed and ambient temperature on the key parameters and the dynamic characteristics of the key parameters.

[0154] The construction module 300 is used to establish the correlation between the wind speed change rate and the change characteristics of multiple key parameters, and establish an abnormal identification model for deteriorated components. Using the change characteristics of multiple key parameters, the correlation, and the change characteristic analysis data as training data, train the abnormal identification model for deteriorated components.

[0155] The detection module 400 is used to obtain the actual value of the component to be detected and the predicted value output by the trained abnormal identification model for deteriorated components, calculate the residual index and the abnormal point rate according to the actual value and the predicted value, and when it is determined that the component to be detected is abnormal according to the residual index and the abnormal point rate, perform the deterioration warning process on the component to be detected.

[0156] It should be noted that the above explanation of the embodiments of the warning method for the deterioration of wind turbine components also applies to the device of this embodiment. The principles of each module to implement its functions are the same and will not be elaborated here.

[0157] In summary, the warning device for the deterioration of wind turbine components in the embodiments of the present application constructs a deteriorated component identification model by combining data mining and thermal characteristic analysis, analyzes the deterioration of key systems and key equipment of the wind turbine, and can accurately locate the deteriorated trend components before the occurrence of faults, improving the accuracy and pertinence of component deterioration identification. It is beneficial to timely discover the abnormalities of the wind turbine and perform the operation processing of the fan in advance, which is beneficial to ensuring the stability and safety of the operation of the wind turbine.

[0158] To implement the above embodiments, the present application also proposes an electronic device, which includes: a processor; a memory for storing processor-executable instructions; wherein, the processor is configured to execute the instructions to implement the warning method for the deterioration of wind turbine components as described in any one of the embodiments of the first aspect above.

[0159] To implement the above embodiments, the present application also provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements the early warning method for the deterioration of wind turbine components as described in any one of the embodiments of the first aspect above.

[0160] In the description of this specification, the description with reference to terms such as "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described can be combined in a suitable manner in any one or more embodiments or examples. In addition, without contradiction, those skilled in the art can combine and combine the different embodiments or examples described in this specification and the features of different embodiments or examples.

[0161] In addition, the terms "first" and "second" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, the features defined with "first" and "second" can explicitly or implicitly include at least one of the features. In the description of the present application, "a plurality" means at least two, such as two, three, etc., unless otherwise specifically and clearly defined.

[0162] Any process or method description in a flowchart or described in other ways herein can be understood to represent a module, segment, or portion of code including one or more executable instructions for implementing a customized logic function or process. The scope of the preferred embodiments of the present application includes additional implementations, where the functions can be executed in a substantially simultaneous manner or in a reverse order according to the functions involved, rather than in the order shown or discussed, which should be understood by those skilled in the art of the embodiments of the present application.

[0163] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a definable sequence list of executable instructions for implementing logical functions, and can be specifically implemented in any computer-readable medium for use by an instruction execution system, apparatus, or device (such as a computer-based system, a system including a processor, or other systems that can fetch instructions from the instruction execution system, apparatus, or device and execute the instructions), or in combination with these instruction execution systems, apparatuses, or devices. For the purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by or in combination with an instruction execution system, apparatus, or device. More specific examples (non-exhaustive list) of computer-readable media include the following: an electrical connection portion with one or more wirings (electronic device), a portable computer diskette (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). Additionally, the computer-readable medium can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other media, followed by editing, interpretation, or other suitable processing as necessary, and then storing it in a computer memory.

[0164] It should be understood that various parts of the present application can be implemented by hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, any one or a combination of the following techniques well known in the art can be used: discrete logic circuits having logic gate circuits for implementing logical functions on data signals, application-specific integrated circuits having suitable combinational logic gate circuits, programmable gate arrays (PGAs), field programmable gate arrays (FPGAs), etc.

[0165] Those of ordinary skill in the art of this technology can understand that all or part of the steps carried by the method of the above embodiments can be completed by instructing relevant hardware through a program, and the program can be stored in a computer-readable storage medium. When the program is executed, it includes one or a combination of the steps of the method embodiments.

[0166] In addition, each functional unit in various embodiments of the present application may be integrated into a processing module, may exist separately physically for each unit, or two or more units may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of a software functional module. When the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may also be stored in a computer-readable storage medium.

[0167] The above-mentioned storage medium may be a read-only memory, a magnetic disk, an optical disc, etc. Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and should not be construed as limiting the present application. Those of ordinary skill in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of the present application.

Claims

1. A warning method for the deterioration of wind turbine components, characterized in that, It includes the following steps: Analyze the thermal characteristics of multiple key components in a wind turbine during the thermal equilibrium process to determine multiple key parameters related to the wind speed and an abnormal identification mechanism based on the multiple key parameters; According to the abnormal identification mechanism, analyze the change characteristics of the multiple key parameters by combining the influence of the wind speed and ambient temperature on the key parameters and the dynamic characteristics of the key parameters; Construct the correlation relationship between the wind speed change rate and the change characteristics of the multiple key parameters, and construct an abnormal identification model for deteriorated components. Use the change characteristics of the multiple key parameters, the correlation relationship, and the change characteristic analysis data as training data to train the abnormal identification model for deteriorated components; Obtain the actual value of the component to be detected and the predicted value output by the trained abnormal identification model for deteriorated components, calculate the residual index and the abnormal point rate based on the actual value and the predicted value. When it is determined that the component to be detected is abnormal according to the residual index and the abnormal point rate, perform a deterioration warning process on the component to be detected.

2. The method according to claim 1, characterized in that After performing the deterioration warning process on the component to be detected, it further includes: Analyze the abnormal parameters of the component to be detected through the Fault Tree Analysis (FTA) algorithm to determine the abnormal cause; Analyze the abnormal cause and the abnormal parameters of the component to be detected through the Failure Mode and Effects Analysis (FMEA) algorithm to obtain an abnormal diagnosis knowledge table, where the abnormal diagnosis knowledge table includes the abnormal events that occur to the component to be detected; Perform operation and maintenance on the component to be detected according to the abnormal diagnosis knowledge table.

3. The method according to claim 1, characterized in that, The multiple key parameters include active power and multiple temperature parameters. The multiple temperature parameters include the temperatures of different units of each key component. Analyzing the change characteristics of the multiple key parameters includes: Determine the Pearson correlation between the wind speed and the temperature parameter according to the temperature values of different units of each key component at different wind speeds; Analyze the proportion of different temperature values in each wind speed interval and the change law of different temperature values with the wind speed according to the distribution of the values of the multiple temperature parameters in different wind speed intervals; Analyze the proportion of different temperature values in each ambient temperature interval and the change law of different temperature values with the ambient temperature according to the distribution of the values of the multiple temperature parameters in different ambient temperature intervals; Based on the dynamic fluctuation characteristics of the temperature parameter, set the time interval for parameter change, construct a box plot of the temperature parameter change according to the time interval, and analyze the change trend of the multiple temperature parameters over a long period through the box plot.

4. The method according to claim 3, characterized in that, Analyzing the change characteristics of the multiple key parameters further includes: Determine the maximum and minimum values of the temperature parameter within a preset sliding window according to the distribution of the temperature parameter; Obtain multiple extreme points of the temperature parameter within the preset sliding window by calculating local extreme points; Calculate the change rate of the temperature parameter within the preset sliding window using the maximum value, the minimum value, and the multiple extreme points.

5. The method according to claim 4, wherein The method of obtaining multiple extreme points of the temperature parameter within the preset sliding window by calculating local extreme points includes: Smoothing the temperature parameter and the wind speed parameter within the preset sliding window through the Savitzky-Golay data smoothing algorithm; Analyzing the number of maximum values under different smoothing parameters, and respectively determining the smoothing index and the width of local extreme points corresponding to the temperature parameter and the wind speed parameter; Processing the temperature parameter and the wind speed parameter respectively according to the corresponding smoothing index and the width of local extreme points, and analyzing the frequency distribution of the distances between the obtained extreme points.

6. The method according to claim 1, wherein The construction of the abnormal component deterioration recognition model includes: For each of the key components, constructing a corresponding abnormal recognition model for each of the key parameters; Among them, constructing a corresponding abnormal recognition model for each of the key parameters includes: Constructing a temperature rise abnormal recognition model and a temperature drop abnormal recognition model for each of the key parameters respectively according to two situations where the temperature rate is positive and the temperature rate is negative.

7. The method according to claim 1, characterized in that The calculation of the residual index and the abnormal point rate according to the actual value and the predicted value includes: In the case where the actual value is greater than the predicted value, calculating the difference between the actual value and the predicted value, and determining the abnormal threshold according to the residual distribution of the actual value and the predicted value; Calculating the residual index according to the difference and the abnormal threshold, where the residual index represents the abnormal degree of the parameter; Calculating the abnormal point rate according to the sliding window of the abnormal component deterioration recognition model and the number of abnormal points within the sliding window.

8. An early warning device for the deterioration of wind turbine components, characterized in that, It includes the following modules: A thermal analysis module for analyzing the thermal characteristics of multiple key components in a wind turbine during the thermal balance process, determining multiple key parameters related to the wind speed, and the abnormal recognition mechanism based on the multiple key parameters; A parameter change analysis module for analyzing the change characteristics of the multiple key parameters according to the abnormal recognition mechanism, combining the influence of the wind speed and the ambient temperature on the key parameters and the dynamic characteristics of the key parameters; A construction module for constructing the correlation between the wind speed change rate and the change characteristics of the multiple key parameters, constructing an abnormal component deterioration recognition model, and using the change characteristics of the multiple key parameters, the correlation, and the change characteristic analysis data as training data to train the abnormal component deterioration recognition model; A detection module for obtaining the actual value of the component to be detected and the predicted value output by the trained abnormal component deterioration recognition model, calculating the residual index and the abnormal point rate according to the actual value and the predicted value, and performing a deterioration early warning process on the component to be detected when it is determined that the component to be detected is abnormal according to the residual index and the abnormal point rate.

9. An electronic device, including: A processor; A memory for storing instructions executable by the processor; Wherein, the processor is configured to execute the instructions to implement the early warning method for the deterioration of wind turbine components as described in any one of claims 1-7.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the early warning method for the deterioration of wind turbine components described in any one of claims 1-7.

Images