Operation fault monitoring and early warning method and system suitable for wind generating set

Abstract

The invention relates to the technical field of wind generating sets, in particular to an operation fault monitoring and early warning method and system suitable for a wind generating set, and the method specifically comprises the steps: constructing an abnormal significant coefficient of a part in each time period through the waveform characteristics of a vibration signal spectrogram of the wind generating set and the comparison with a normal vibration spectrogram; constructing an abnormal vibration coupling index of each time period based on the difference between the abnormal saliency coefficients of all the components in each time period and the overall distribution size; fitting a wind speed power curve based on data points composed of wind speed and power data at each moment; analyzing the synchronism of the distance between each data point and the wind speed power curve and the abnormal vibration coupling index, and constructing an abnormal synchronization coefficient; performing fault monitoring on the wind generating set based on the abnormal synchronization coefficient; the influence of the dynamic load on the operation fault monitoring of the generator set is avoided, the operation fault of the generator set can be found in time, and the fault monitoring accuracy of the wind generating set is improved.

Description

Technical Field

[0001] This application relates to the technical field of wind turbines, and particularly to an operation fault monitoring and early warning method and system applicable to wind turbines. Background Art

[0002] Wind power generation has occupied a place in the new energy field due to its high efficiency, environmental protection and strong sustainability. Limited by the distribution of wind energy, most wind farms are located in areas with relatively harsh natural environments. Coupled with the complex mechanical structure of wind turbines, each component is extremely vulnerable to damage. Therefore, it is very necessary to monitor the status and early warn of faults of wind turbines.

[0003] A wind turbine consists of multiple complex components, and there are energy flows and coupling relationships between the components, which further increases the complexity and difficulty of fault monitoring. During long-term operation, complex climatic conditions are likely to cause the unit to be subjected to uneven dynamic loads, increasing the vibration and fatigue damage of the unit. Existing methods usually fail to fully consider the impact of dynamic loads on the operation faults of the generator set, and it is difficult to accurately reflect the fault operation state of the generator set in a timely manner, resulting in insufficient fault early warning ability. Summary of the Invention

[0004] In order to solve the above technical problems, the purpose of this application is to provide an operation fault monitoring and early warning method and system applicable to wind turbines. The specific technical solutions adopted are as follows: In the first aspect, an embodiment of this application provides an operation fault monitoring and early warning method applicable to wind turbines. The method includes the following steps: For each acquisition moment of the wind speed and power of the wind turbine, within the time period between adjacent acquisition moments, collect the vibration data of each component in the wind turbine, including the vibration data of the main shaft, gearbox and generator shaft; Record the fitting curve of the vibration data of the component in each time period as the vibration fitting curve; construct the impact intensity coefficient of each impact signal in the vibration fitting curve based on the prominence of the wave peaks of each impact signal in the vibration fitting curve; construct the distribution law coefficient of the impact signals in the vibration fitting curve based on the time distribution difference of the peak points of the impact signals in the vibration fitting curve, and combine the distribution law coefficient to construct the abnormal impact coefficient of the component in each time period; Compare the spectrogram of the vibration data time series of the component in each time period with the spectrogram of the historical normal vibration data to construct the spectral difference sequence of the component in each time period; construct the spectral difference anomaly value of the component in each time period based on the change characteristics and overall data size of the spectral difference sequence; Construct the anomaly significance coefficient of components in each time period based on the anomaly impact coefficient and the spectrum difference anomaly value; construct the anomaly vibration coupling index of each time period based on the difference between the anomaly significance coefficients of all components in each time period and the overall distribution size. Fit the wind speed-power curve based on the data points composed of wind speed and power data at each moment; construct the anomaly synchronization coefficient of the wind turbine based on the distance between each data point and the wind speed-power curve and the anomaly vibration coupling index, and perform fault monitoring of the wind turbine based on the anomaly synchronization coefficient.

[0005] In one embodiment, the process of obtaining the impact intensity coefficient is as follows: Obtain the peak points of the impact signals in the vibration fitting curve through the peak search algorithm; obtain the peak widths of the wave peaks where each peak point is located; take the product of the amplitude of the peak point of each impact signal and the peak width of the wave peak where it is located as the impact intensity coefficient of each impact signal.

[0006] In one embodiment, the process of constructing the distribution law coefficient of the impact signals in the vibration fitting curve and constructing the anomaly impact coefficient of components in each time period in combination with the distribution law coefficient is as follows: Calculate the difference amount between the corresponding times of the peak points of two adjacent impact signals in the vibration fitting curve, and record the standard deviation of all the difference amounts of the vibration fitting curve as the distribution law coefficient; Record the anomaly impact coefficient of the vibration data of the main shaft in the i-th time period as , The expression of is: , where, is the mean value of all the impact intensity coefficients of the main shaft vibration fitting curve in the i-th time period; is the distribution law coefficient of the impact signals in the main shaft vibration fitting curve in the i-th time period; is a preset extremely small positive number.

[0007] In one embodiment, the process of obtaining the spectrum difference sequence is as follows: Obtain the spectrogram of the time series of the vibration data of the components in each time period, denoted as the first spectrogram; obtain the spectrogram of the time series of the historical normal vibration data of the components with the same time length as each time period, denoted as the second spectrogram; record the sequence composed of the differences of the amplitudes of all the same frequencies between the first spectrogram and the second spectrogram as the spectrum difference sequence of the components in each time period.

[0008] In one embodiment, the process of obtaining the spectrum difference anomaly value of the components in each time period is as follows: Adopt the trend test algorithm to obtain the test statistic of the trend change of the spectrum difference sequence; Calculate the result of the exponential function with the base of the natural constant and the test statistic of the trend change of the spectral difference sequence of the components at each time period as the exponent; take the product of the mean of all data in the spectral difference sequence of the components at each time period and the calculated result as the spectral difference outlier of the vibration data of the components at each time period.

[0009] In one embodiment, the anomaly significance coefficient of the components at each time period is: the product of the anomaly impact coefficient of the components at each time period and the spectral difference outlier.

[0010] In one embodiment, the process of obtaining the abnormal vibration coupling index of each time period is as follows: Calculate the difference between the anomaly significance coefficients of any two components at each time period, denoted as the first difference; Take the ratio of the mean of the anomaly significance coefficients of all components within each time period to the mean of all the first differences as the abnormal vibration coupling index of each time period.

[0011] In one embodiment, the process of obtaining the wind speed power curve is as follows: Taking the wind speed as the abscissa and the power as the ordinate, construct a wind speed-power scatter plot through the wind speed and power data collected at all times; use the curve fitting algorithm to obtain the fitting curve for the data points with non-zero power in the wind speed-power scatter plot, denoted as the wind speed power curve.

[0012] In one embodiment, constructing the abnormal synchronization coefficient of the wind turbine generator set and performing fault monitoring on the wind turbine generator set based on the abnormal synchronization coefficient is specifically as follows: For the data points with non-zero power in the wind speed-power scatter plot, calculate the shortest distance between each data point and the wind speed power curve as the wind speed power state offset of each data point; Denote the sequence composed of the abnormal vibration coupling indices of all time periods as the abnormal vibration coupling index sequence; denote the sequence composed of the wind speed power state offsets of all data points as the wind speed power state offset sequence; denote the metric distance between the abnormal vibration coupling index sequence and the wind speed power state offset sequence as the abnormal synchronization coefficient; If the normalized value of the abnormal synchronization coefficient is less than or equal to the preset fault warning threshold, it indicates that there is a fault in the operation of the wind turbine generator set; otherwise, it indicates that the wind turbine generator set is operating normally.

[0013] In a second aspect, an embodiment of the present application further provides an operation fault monitoring and warning system applicable to a wind turbine generator set, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of the method described in any one of the above.

[0014] The embodiments of the present application have at least the following beneficial effects: The present application provides an operation fault monitoring and early warning method and system applicable to a wind turbine generator set. By collecting the wind speed and power data of the wind turbine generator set at each moment, within the time period between adjacent collection moments, the vibration data of each component in the wind turbine generator set is collected, including the vibration data of the main shaft, gearbox, and generator shaft. By analyzing the waveform characteristics of the spectrogram of the vibration signal of each component when a fault occurs in the wind turbine generator set and comparing it with the spectrogram of the normal vibration data, the change characteristics and overall data size of the spectral difference sequence are analyzed, and the abnormal significance coefficient of each component in each time period is constructed, deeply analyzing the abnormal impact signal and spectral abnormal difference characteristics that may occur due to abnormal vibration in different components of the wind turbine generator set. Based on the differences and overall distribution sizes between the abnormal significance coefficients of all components in each time period, the abnormal vibration coupling index of each time period is constructed, considering the vibration coupling relationship between different components, and reducing the interference of abnormal vibration between components on fault monitoring. Based on the data points composed of the wind speed and power data at each moment, the wind speed-power curve is fitted. By analyzing the synchronism between the distance between each data point and the wind speed-power curve and the abnormal vibration coupling index, the abnormal synchronization coefficient of the wind turbine generator set is constructed, which can accurately evaluate the possibility of an abnormal state in the operation of the wind turbine generator set under the influence of dynamic loads. Based on the abnormal synchronization coefficient, the fault monitoring of the wind turbine generator set is carried out, avoiding the influence of dynamic loads on the operation fault monitoring of the generator set, helping to timely detect the operation faults of the generator set, and improving the accuracy of the wind turbine generator set fault monitoring. Description of the Drawings

[0015] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0016] Figure 1 It is a flowchart of the steps of an operation fault monitoring and early warning method applicable to a wind turbine generator set provided by an embodiment of the present application; Figure 2 It is a schematic diagram of the acquisition process of the abnormal vibration coupling index. Detailed Embodiments

[0017] To further elaborate on the technical means and effects adopted by this application to achieve the intended invention purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation manners, structures, features, and effects of the operation fault monitoring and early warning method and system applicable to wind turbines proposed according to this application. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. In addition, the specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0018] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the technical field to which this application belongs.

[0019] The following specifically describes the specific solutions of the operation fault monitoring and early warning method and system applicable to wind turbines provided by this application in conjunction with the accompanying drawings.

[0020] Please refer to Figure 1 , which shows the step flowchart of the operation fault monitoring and early warning method applicable to wind turbines provided by one embodiment of this application. The method includes the following steps: Step S1, for each acquisition moment of the wind speed and power of the wind turbine, within the time period between adjacent acquisition moments, collect the vibration data of each component in the wind turbine, including the vibration data of the main shaft, gearbox, and generator shaft.

[0021] The probability of occurrence of vibration faults in wind turbines is the highest, the amount of information contained in vibration signals is the largest and the real-time performance is good. During the operation of the wind turbine, synchronously collect the vibration data of the main shaft, gearbox, and generator shaft in the wind turbine respectively. Preferably, in the embodiment of this application, the acquisition frequency of the vibration data is set to 1000HZ. As other embodiments of this application, implementers can set the acquisition frequency of the vibration data according to the actual situation.

[0022] At the same time, collect the wind speed and power data of the wind turbine. Preferably, in the embodiment of this application, the acquisition time intervals of the wind speed and power data of the wind turbine are both set to 5 seconds. As other embodiments of this application, implementers can set the acquisition time intervals of the wind speed and power data according to the actual situation.

[0023] For each acquisition moment of the wind speed and power data, use the time period between this acquisition moment and the previous acquisition moment as the time period corresponding to this acquisition moment, and the wind speed and power data at each acquisition moment correspond to the vibration data collected within its corresponding time period.

[0024] Among them, the vibration data, wind speed data, and power data are all collected through the SCADA system of the wind turbine generator set. The SCADA system is provided as a standard by the manufacturer of the wind turbine generator set and can collect and process data from various sensors in real time to monitor and manage the operating status of the unit.

[0025] Regarding the collection duration of all data, preferably, in the embodiments of the present application, the collection duration is set to 12 hours. As other embodiments of the present application, the implementer can set the collection duration according to the actual situation.

[0026] Step S2, record the fitting curve of the vibration data of the component in each time period as the vibration fitting curve; construct the impact intensity coefficient of each impact signal in the vibration fitting curve based on the prominence of the wave peaks of each impact signal in the vibration fitting curve; construct the distribution law coefficient of the impact signals in the vibration fitting curve based on the time distribution difference of the peak points of the impact signals in the vibration fitting curve, and combine the distribution law coefficient to construct the abnormal impact coefficient of the component in each time period.

[0027] The wind turbine generator set uses the impeller system to capture natural wind energy and convert it into mechanical energy. When the wind blows the blades to rotate, the kinetic energy is transmitted through the transmission system and finally drives the generator to convert it into electrical energy. The transmission efficiency of its kinetic energy directly affects the power generation status of the wind turbine generator set. Since there are bearing friction losses inside the main shaft, gearbox, and generator shaft, when there are foreign objects in the bearing or the lubrication is poor, the friction loss will increase, and the irregular vibration generated by the wear will affect the normal operation of the generator set, and further interfere with the stable output of the power of the generator set under the influence of dynamic loads. Dynamic loads refer to the continuous change of the force state of the wind turbine generator set caused by climate condition factors such as wind speed and wind direction. Dynamic loads may cause fatigue damage, increased vibration, and other safety hazards to each component in the wind turbine generator set.

[0028] Taking the bearing vibration inside the main shaft as an example, during normal operation, the amplitude of the vibration data is small and basically a stable random waveform. However, in the case of poor lubrication, the lubricating oil cannot form an effective oil film, resulting in an increase in internal friction of the bearing; or, when the load borne is too large, the internal stress of the bearing may exceed its bearing limit, thus causing deformation of the rolling elements or raceways. These factors may all cause abnormal vibrations to occur. Compared with the time-domain waveform of the vibration data during normal operation, the time-domain waveform of the vibration data after the abnormality has obvious impact signals, and the overall kurtosis value of the vibration data will also increase significantly. And under the action of the periodic rotation of the bearing, the appearance of the impact signals has strong regularity. Therefore, the following analysis and processing are performed on the vibration data.

[0029] Taking the vibration data of the main shaft in the i-th time period as an example, since there are obvious amplitude changes in the vibration data at the impact signal position, first, all the vibration data of the main shaft in this time period are used as the input of the least squares method for curve fitting, and the obtained fitting curve is denoted as the vibration fitting curve of the main shaft in this time period; then, the automatic multi-scale peak search algorithm (AMPD) is used to obtain the peak points of the impact signals in the vibration fitting curve. Among them, the least squares method and the automatic multi-scale peak search algorithm are both well-known technologies, and the specific processes will not be elaborated.

[0030] Secondly, obtain the peak widths of the wave peaks where the peak points of each impact signal are located. Among them, the process of obtaining the peak widths is a well-known technology, and the specific process will not be elaborated; calculate the product of the amplitude of the peak point of each impact signal and the peak width of the wave peak where it is located as the impact intensity coefficient of each impact signal; calculate the mean value of all the impact intensity coefficients of the vibration fitting curve of the main shaft in the i-th time period, denoted as The impact intensity coefficient reflects the degree characteristics of abnormal vibration. The higher the peak value of the impact signal and the longer the action time of the impact signal, the greater the influence of the existing abnormal vibration, and the greater the impact intensity coefficient.

[0031] Furthermore, since the distribution of impact signals in the vibration data has strong regularity, therefore, calculate the absolute value of the difference between the corresponding times of the peak points of two adjacent impact signals in the vibration fitting curve, denoted as the first time difference, and take the standard deviation of all the first time differences as the distribution regularity coefficient of the impact signals in the vibration fitting curve. The smaller the distribution regularity coefficient, the more regular the appearance of the impact signals in the vibration data.

[0032] Finally, calculate the abnormal impact coefficient of the vibration data of the main shaft in each time period, and the expression is: , where is the abnormal impact coefficient of the vibration data of the main shaft in the i-th time period; is the mean value of all the impact intensity coefficients of the vibration fitting curve of the main shaft in the i-th time period; is the distribution regularity coefficient of the impact signals in the vibration fitting curve of the main shaft in the i-th time period; is a preset extremely small positive number, and its function is to avoid the denominator being zero. Preferably, in the embodiments of the present application, is set to 0.1. As other embodiments of the present application, the implementer can set the value of according to the actual situation.

[0033] The larger the

[0034] Step S3: Compare the spectrograms of the vibration data time series of the components in each time period with the spectrograms of the historical normal vibration data, and construct the spectrogram difference sequences of the components in each time period; based on the change characteristics and overall data size of the spectrogram difference sequences, construct the spectrogram difference outliers of the components in each time period.

[0035] When a main shaft operation fault occurs in a generator set, the partial frequency amplitudes of the vibration data spectrogram will suddenly increase. There are significant differences between the spectral data corresponding to the abnormal vibration data and the normal vibration data, and the corresponding spectral difference curve becomes larger and more unstable as the frequency increases.

[0036] Take the time series of the main shaft vibration data in the i-th time period as the input of the discrete Fourier transform algorithm, and denote the output spectrogram as the first spectrogram; take the time series of the historical normal vibration data of the main shaft with the same time length as the input of the discrete Fourier transform algorithm, and denote the obtained spectrogram as the second spectrogram; obtain the sequence composed of the differences of all the same frequency amplitudes between the first spectrogram and the second spectrogram, and denote it as the spectrogram difference sequence of the main shaft in the i-th time period. Use the Mann-Kendall trend test algorithm to analyze the change characteristics of the spectrogram difference sequence, and obtain the test statistic of the trend change of the spectrogram difference sequence, denoted as . The obtained If the symbol is positive and the absolute value is larger, it means that the elements in the spectrogram difference sequence will increase as the frequency increases. Among them, the discrete Fourier transform and the Mann-Kendall trend test algorithm are both well-known technologies, and the specific processes will not be elaborated here.

[0037] Furthermore, calculate the spectrogram difference outliers of the vibration data in each time period. The expression is: , where is the spectrogram difference outlier of the vibration data of the main shaft in the i-th time period; is the mean value of all the data in the spectrogram difference sequence of the main shaft in the i-th time period; is the test statistic of the trend change of the spectrogram difference sequence of the main shaft in the i-th time period; represents the exponential function with the natural constant e as the base.

[0038] The larger is, the greater the difference between the spectrogram of the main shaft vibration data in this time period and the spectrogram of the normal main shaft vibration data; the larger

[0039] Step S4: Construct the anomaly significance coefficient of components in each time period based on the anomaly impact coefficient and the spectrum difference anomaly value; construct the anomaly vibration coupling index of each time period based on the differences and overall distribution sizes among the anomaly significance coefficients of all components in each time period.

[0040] (1) Calculate the anomaly significance coefficient of the main shaft in each time period, and the expression is: , where in the formula, is the anomaly significance coefficient of the main shaft in the i-th time period; is the anomaly impact coefficient of the vibration data of the main shaft in the i-th time period; is the spectrum difference anomaly value of the vibration data of the main shaft in the i-th time period.

[0041] The obtained The larger it is, the more obvious the anomaly impact signal and spectrum anomaly difference of the corresponding vibration data of the main shaft in this time period, and the more likely there is an operating fault.

[0042] (2) Based on the vibration data of the gearbox and the generator shaft in each time period respectively, adopt the same calculation method of the anomaly significance coefficient of the main shaft in each time period to obtain the anomaly significance coefficients of the gearbox and the generator shaft in each time period.

[0043] (3) Since the kinetic energy transfer between generator sets affects each other, and the more serious the coupling vibration effect, the larger the overall anomaly significance coefficients of each component and the smaller the differences between components. To comprehensively evaluate the overall vibration state of the generator set, calculate the absolute value of the difference between the anomaly significance coefficients of any two components in the i-th time period, denoted as the first difference; Take the ratio of the mean value of the anomaly significance coefficients of all components in the i-th time period to the mean value of all the first differences as the anomaly vibration coupling index of the i-th time period. Denote it as . The obtained reflects the overall anomaly vibration state and the coupling characteristics of anomaly vibration among components in the wind turbine generator set.

[0044] Step S5: Fit the wind speed-power curve based on the data points composed of wind speed and power data at each moment; construct the anomaly synchronization coefficient of the wind turbine generator set based on the distances between each data point and the wind speed-power curve and the anomaly vibration coupling index, and conduct fault monitoring of the wind turbine generator set based on the anomaly synchronization coefficient.

[0045] Under the influence of dynamic loads, it may cause fatigue damage, increased vibration and other safety hazards to each component in the wind turbine generator set, and make the wind speed-power data deviate from the normal state. For this reason, within the data acquisition duration, with the wind speed as the abscissa and the power as the ordinate, construct a wind speed-power scatter distribution map through the wind speed and power data collected at all moments.

[0046] Ideally, the overall power generation performance increases with the increase of wind speed. During actual operation, the scatter distribution state of wind speed-power is not a simple linear relationship, but is affected by external factors. When the wind speed changes rapidly, due to the inertial effect of blade rotation and the dynamic response characteristics of the control system, there will be an asynchronous phenomenon between the output power of the unit and the wind speed. When the wind speed suddenly increases, the rotational speed of the impeller cannot immediately keep up with the increase of the wind speed, resulting in that although the wind speed is very high, the power does not reach the theoretical maximum value; on the contrary, when the wind speed drops suddenly, the impeller continues to maintain a high rotational speed due to its own moment of inertia, and at this time, there may be a high power output at a low wind speed. It can be seen that the greater the influence of dynamic loads, the greater the abnormal vibration coupling index, and the more the corresponding wind speed-power data deviates from the ideal distribution area.

[0047] In this application, first, the data points with a power of 0 in the wind speed-power scatter distribution diagram are removed. These data points correspond to the time when the wind turbine generator set is in a shutdown state.

[0048] Then, polynomial fitting is used to obtain the fitting curve of the wind speed-power scatter distribution diagram, denoted as the wind speed-power curve. This curve can reflect the power output characteristics at different wind speeds, and the faults or abnormal conditions of the generator set can be reflected through its data changes. Among them, polynomial fitting is a well-known technology, and the specific process will not be elaborated.

[0049] It should be noted that for the curve fitting of the wind speed-power scatter distribution diagram, this application only provides a curve fitting method. There are many existing curve fitting methods, and implementers can also use other curve fitting algorithms to obtain the fitting curve of the wind speed-power scatter distribution diagram. This application does not make specific restrictions.

[0050] After that, for the data points with a power not equal to 0 in the wind speed-power scatter distribution diagram, calculate the shortest distance between each data point and the wind speed-power curve, and use it as the wind speed-power state offset of this data point.

[0051] Since the higher the abnormal vibration coupling index in each period, the more the corresponding wind speed-power data points deviate from the ideal distribution area. Therefore, arrange the abnormal vibration coupling indices of all periods in ascending order of time, and the formed sequence is denoted as the abnormal vibration coupling index sequence; arrange the wind speed-power state offsets of all data points in ascending order of time, and the formed sequence is denoted as the wind speed-power state offset sequence; calculate the DTW distance between the abnormal vibration coupling index sequence and the wind speed-power state offset sequence, and use the DTW distance as the abnormal synchronization coefficient of the vibration abnormality and the wind speed-power offset under the influence of the dynamic load of the generator set. This value reflects the possibility of faults in the wind turbine generator set. Among them, the calculation of the DTW distance is a well-known technology, and the specific process will not be elaborated.

[0052] By deeply analyzing the abnormal impact signals and spectral anomaly difference characteristics that may occur due to abnormal vibration in different components of the generator set, further considering the coupling relationship of vibrations between different components and the synchronization characteristics with the wind speed power state deviation characteristics, this application calculates the abnormal synchronization coefficient between vibration anomalies and wind speed power deviation. The smaller the obtained abnormal synchronization coefficient, the higher the possibility of a fault in the wind turbine generator set. For quantitative evaluation, this application uses the tanh function to normalize the abnormal synchronization coefficient to obtain the normalized value of the abnormal synchronization coefficient.

[0053] Set the fault warning threshold , preferably, in the embodiment of this application, the value is set to 0.2. As other embodiments of this application, the implementer can set the value according to the actual situation. If the normalized value of the abnormal synchronization coefficient is less than or equal to , it is determined that there is a fault in the operation of the wind turbine generator set and an alarm is given; otherwise, it indicates that the wind turbine generator set is operating normally. Thus, the monitoring of the operation fault of the wind turbine generator set is realized.

[0054] The schematic diagram of the acquisition process of the abnormal vibration coupling index is as Figure 2 shown.

[0055] Based on the same inventive concept as the above method, the embodiment of this application also provides an operation fault monitoring and warning system applicable to a wind turbine generator set, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of any one of the above methods for the operation fault monitoring and warning method applicable to a wind turbine generator set.

[0056] In summary, the embodiment of the present application provides an operation fault monitoring and early warning method applicable to a wind turbine generator set. By collecting the wind speed and power data of the wind turbine generator set at each moment, within the time period between adjacent collection moments, the vibration data of each component in the wind turbine generator set is collected, including the vibration data of the main shaft, gearbox, and generator shaft; by analyzing the waveform characteristics of the frequency spectrum diagrams of the vibration signals of each component when the wind turbine generator set fails, and comparing with the frequency spectrum diagrams of the normal vibration data, the change characteristics of the frequency spectrum difference sequence and the overall data size are analyzed, and the abnormal significance coefficient of each component in each time period is constructed, deeply analyzing the abnormal impact signals and frequency spectrum abnormal difference characteristics that may occur due to abnormal vibration in different components of the wind turbine generator set; based on the differences and overall distribution sizes between the abnormal significance coefficients of all components in each time period, the abnormal vibration coupling index of each time period is constructed, considering the vibration coupling relationship between different components, reducing the interference of abnormal vibration between components on fault monitoring; fitting the wind speed-power curve based on the data points composed of the wind speed and power data at each moment; analyzing the synchronism between the distance between each data point and the wind speed-power curve and the abnormal vibration coupling index, constructing the abnormal synchronization coefficient of the wind turbine generator set, which can accurately evaluate the possibility of abnormal operation of the wind turbine generator set under the influence of dynamic loads; based on the abnormal synchronization coefficient, fault monitoring of the wind turbine generator set is carried out, avoiding the influence of dynamic loads on the operation fault monitoring of the generator set, helping to timely detect the operation faults of the generator set, and improving the accuracy of fault monitoring of the wind turbine generator set.

[0057] It should be noted that the above sequence of embodiments of the present application is only for description and does not represent the superiority or inferiority of the embodiments. And the above specific embodiments of the present application have been described. In addition, the processes depicted in the 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.

[0058] The embodiments in the present application are all described in a progressive manner. The same or similar parts between the embodiments can be referred to each other, and the key points of each embodiment are the differences from other embodiments.

[0059] The above is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for monitoring and warning of operating faults applicable to wind turbine generators, characterized in that, The method includes the following steps: For each acquisition moment of the wind speed and power of the wind turbine generator set, within the time period between adjacent acquisition moments, collect the vibration data of each component in the wind turbine generator set, including the vibration data of the main shaft, gearbox, and generator shaft; Denote the fitting curve of the vibration data of the components in each time period as the vibration fitting curve; construct the impact intensity coefficient of each impact signal in the vibration fitting curve based on the prominence of the wave peaks of each impact signal in the vibration fitting curve; construct the distribution law coefficient of the impact signals in the vibration fitting curve based on the time distribution difference of the peak points of the impact signals in the vibration fitting curve, and combine the impact intensity coefficient to construct the abnormal impact coefficient of the components in each time period; Compare the spectrogram of the vibration data time series of the components in each time period with the spectrogram of the historical normal vibration data to construct the spectral difference sequence of the components in each time period; construct the spectral difference outlier of the components in each time period based on the change characteristics and overall data size of the spectral difference sequence; Construct the abnormal significance coefficient of the components in each time period based on the abnormal impact coefficient and the spectral difference outlier; construct the abnormal vibration coupling index of each time period based on the difference and overall distribution size between the abnormal significance coefficients of all components in each time period; Fit the wind speed-power curve based on the data points composed of the wind speed and power data at each moment; construct the abnormal synchronization coefficient of the wind turbine generator set based on the distance between each data point and the wind speed-power curve and the abnormal vibration coupling index, and perform fault monitoring of the wind turbine generator set based on the abnormal synchronization coefficient.

2. The operation fault monitoring and early warning method applicable to a wind power generating set according to claim 1, wherein, The process of obtaining the impact intensity coefficient is as follows: Obtain the peak points of the impact signals in the vibration fitting curve through the peak search algorithm; obtain the peak width of the wave peaks where each peak point is located; take the product of the amplitude of the peak point of each impact signal and the peak width of the wave peak where it is located as the impact intensity coefficient of each impact signal.

3. The operation fault monitoring and early warning method applicable to a wind power generating set according to claim 1, wherein, The specific process of constructing the distribution law coefficient of the impact signals in the vibration fitting curve and combining the distribution law coefficient to construct the abnormal impact coefficient of the components in each time period is as follows: Calculate the difference amount between the corresponding times of the peak points of two adjacent impact signals in the vibration fitting curve, and denote the standard deviation of all the difference amounts of the vibration fitting curve as the distribution law coefficient; Denote the abnormal impact coefficient of the vibration data of the main shaft in the $i$-th period as , The expression of is as follows: , where is the mean value of all impact intensity coefficients of the spindle vibration fitting curve in the i-th time period; is the distribution law coefficient of the impact signal in the spindle vibration fitting curve in the i-th time period; is a preset extremely small positive number.

4. The operation fault monitoring and early warning method applicable to a wind power generating set according to claim 1, wherein The process of obtaining the spectral difference sequence is as follows: Obtain the spectrogram of the time series of the vibration data of the components in each time period, denoted as the first spectrogram; obtain the spectrogram of the time series of the historical normal vibration data of the components with the same time length as each time period, denoted as the second spectrogram; denote the sequence composed of the differences of the amplitudes of all the same frequencies between the first spectrogram and the second spectrogram as the spectral difference sequence of the components in each time period.

5. The operation fault monitoring and early warning method applicable to a wind power generation set according to claim 1, wherein The process of obtaining the spectral difference outlier of the components in each time period is as follows: Adopt the trend test algorithm to obtain the test statistic of the trend change of the spectral difference sequence; Calculate the calculation result of the exponential function with the natural constant as the base and the test statistic of the trend change of the spectral difference sequence of the components in each time period as the exponent; Take the product of the mean value of all the data in the spectral difference sequence of the components in each time period and the calculation result as the spectral difference outlier of the vibration data of the components in each time period.

6. The operation fault monitoring and early warning method applicable to a wind power generating set according to claim 1, wherein The abnormal significance coefficient of the component at each time period is: the product of the abnormal impact coefficient of the component at each time period and the abnormal value of the spectrum difference.

7. The operation fault monitoring and early warning method applicable to a wind turbine generator set according to claim 1, characterized in that, The process of obtaining the abnormal vibration coupling index at each time period is as follows: Calculate the difference between the abnormal significance coefficients of any two components at each time period, denoted as the first difference; Take the ratio of the mean value of the abnormal significance coefficients of all components within each time period to the mean value of all the first differences as the abnormal vibration coupling index at each time period.

8. The operation fault monitoring and early warning method applicable to a wind power generating set according to claim 1, characterized in that, The process of obtaining the wind speed-power curve is as follows: Taking the wind speed as the abscissa and the power as the ordinate, construct a wind speed-power scatter distribution diagram through the wind speed and power data collected at all times; use the curve fitting algorithm to obtain the fitting curve for the data points with non-zero power in the wind speed-power scatter distribution diagram, denoted as the wind speed-power curve.

9. The operation fault monitoring and early warning method applicable to a wind power generating set according to claim 8, characterized in that, Construct the abnormal synchronization coefficient of the wind turbine generator set, and perform fault monitoring on the wind turbine generator set based on the abnormal synchronization coefficient, specifically: For the data points with non-zero power in the wind speed-power scatter distribution diagram, calculate the shortest distance between each data point and the wind speed-power curve as the wind speed-power state offset of each data point; Denote the sequence composed of the abnormal vibration coupling indices at all time periods as the abnormal vibration coupling index sequence; denote the sequence composed of the wind speed-power state offsets of all data points as the wind speed-power state offset sequence; denote the metric distance between the abnormal vibration coupling index sequence and the wind speed-power state offset sequence as the abnormal synchronization coefficient; If the normalized value of the abnormal synchronization coefficient is less than or equal to the preset fault warning threshold, it indicates that there is a fault in the operation of the wind turbine generator set; otherwise, it indicates that the wind turbine generator set is operating normally.

10. An operation fault monitoring and warning system applicable to a wind turbine generator set, 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, it implements the steps of the method according to any one of claims 1-9.

Images