A gearbox condition monitoring method based on stator current signal of doubly-fed wind turbine

Abstract

The present invention discloses a gearbox state monitoring method based on the stator current signal of a doubly-fed wind turbine. The present invention clarifies the current signal composition through a phenomenological model of the stator current signal, and calculates the state monitoring index based on the model using a current energy ratio algorithm; and uses a splitting and reorganization method to solve the data spectrum overlap and the problem of being unable to uniformly compare and evaluate caused by variable operating conditions; the speed estimation method provides an alternative solution to the same speed source for older models of wind turbines. The present invention proposes a splitting and reorganization method, which intends to use a uniform splitting of the current signal according to the speed label and classifying it within the same speed range to avoid frequency band aliasing and thus solve the problem of being unable to compare variable operating conditions. The present invention achieves a low-cost method for gearbox state monitoring that only requires a current transformer by establishing a phenomenological model, constructing a current energy ratio algorithm index, and proposing a splitting and reorganization method and a speed estimation method.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and more specifically, to a gearbox status monitoring method based on a stator current signal of a doubly-fed wind turbine. Background Art

[0002] At present, the domestic wind power industry is developing rapidly, but most wind turbines are installed at high altitudes in relatively open areas. The working environment is very harsh and they are eroded by wind and sand all year round. The probability of failure is very high. Therefore, it is particularly important to be able to detect in real time and find wind turbine failures as early as possible. At present, wind turbines are mainly detected by two methods: traditional vibration detection method and stator current detection method. In the prior art, the gearbox fault diagnosis principle based on stator current is only carried out under the condition of rotating power generation, not under the state of stator voltage vector control (double-fed grid connection). Although the stator current signal fault characteristics are clear and the fault information is rich, the quality of the generated power does not meet the grid connection requirements.

[0003] The control method of stator voltage vector control (doubly fed grid-connected) will make the wind turbine converge its energy to 50Hz as much as possible regardless of the operating conditions to meet the grid connection requirements. Relatively speaking, the fault information in the stator current signal will be greatly affected, making fault diagnosis more difficult. Summary of the invention

[0004] The present invention provides a gearbox status monitoring method based on the stator current signal of a doubly-fed wind turbine. The method divides the acquired current data and groups and reorganizes the current data according to the speed data as a label to avoid frequency band aliasing and obtain effective information of the stator current signal, thereby solving the problem of difficulty in fault diagnosis under doubly-fed grid-connected conditions.

[0005] The present invention provides a gearbox state monitoring method based on a stator current signal of a doubly-fed wind turbine, comprising:

[0006] Step S1: reading a current signal and a speed signal of a wind turbine generator set in the time domain, and sampling the current signal and the speed signal to obtain current data and speed data;

[0007] Step S2: dividing the current data by zero-crossing detection, wherein the number of the current data divided is equal to the number of the speed data;

[0008] Step S3: matching the rotation speed data with the divided current data;

[0009] Step S4: Based on the rotation speed data as a label, all current data matched by the rotation speed data belonging to the same preset rotation speed range are divided into the same group;

[0010] Step S5: reorganizing a preset number of current data belonging to the same group to obtain reorganized current data of each group;

[0011] Step S6: based on the pre-established phenomenological model of the stator current signal of the doubly-fed wind turbine generator set, respectively calculating the current energy ratio according to the reorganized current data of each group;

[0012] Step S7: Determine the state of the gearbox according to the average value of the current energy ratio obtained in each group.

[0013] Optionally, in step S1, the sampling frequency of the current data is 10000 Hz, the sampling frequency of the rotation speed data is 2 Hz, and the data are collected continuously for 1 hour.

[0014] Optionally, within the speed range of 1545 r / min-1785 r / min, each 30 r / min is a preset speed range.

[0015] Optionally, the preset number in step S5 is 30.

[0016] Optionally, the formula of the model in step S6 is:

[0017] I=a1sin(2×50×π×t)+a2sin(2×erf×π×t)+a3sin(2×(2×(erf-50)+50)×π×t)+

[0018] a4sin(2×(6×(erf-50)+50)×π×t)+a5×sin(2×(6×(erf-50)+5×50)×π×t)+

[0019] a7sin(2×(6×(erf-50)+7×50)×π×t)+a 11 sin(2×(2×6×(erf-50)+11×50)×π×t)+a 13 sin(2×(2×6×(erf-50)+13×50)×π×t)+a 17 sin(2×(3×6×(erf-50)+17×50)×π×t)

[0020] Where I represents the stator current signal, a i represents the modulation amplitude, t represents time, erf represents electromagnetic frequency, and 50Hz represents power frequency;

[0021] The formula for the current energy ratio is:

[0022]

[0023] In the formula, SER represents the current energy ratio, Second sideband energy represents the sideband energy corresponding to a3sin(2×(2×(erf-50)+50)×π×t) in the model, and First sideband energy represents the sideband energy corresponding to a2sin(2×erf×π×t) in the model. The sideband energy represents the sum of the spectrum amplitudes in the corresponding frequency range.

[0024] Optionally, step S7 includes: if the average value of the current-energy ratio is greater than a pre-determined current-energy ratio of a normal state, determining that there is a fault in the state of the gearbox.

[0025] Optionally, the method for acquiring the speed data in step S1 is:

[0026] Divide the current data per minute into 60 parts corresponding to each second;

[0027] Convert the current data per second into a current frequency domain signal through Fourier transform, and obtain the local maximum energy value of the high frequency part of the current frequency domain signal;

[0028] The rotation speed data per second is acquired according to the local maximum value of energy.

[0029] Compared with the prior art, the present invention has the following beneficial effects:

[0030] The present invention can perform fault diagnosis under the state of stator voltage vector control (double-fed grid connection) based on the current energy ratio algorithm. In addition, the present invention can solve the problem of data spectrum overlap and inability to unify comparative evaluation caused by variable working conditions by segmenting and reorganizing the current data, so that the gearbox state monitoring can be achieved through current data and speed data, and the cost is low. In addition, the present invention can also estimate the speed through current data, so that the gearbox state monitoring can be achieved only through current data. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For those skilled in the art, other drawings can be obtained based on these drawings without creative work.

[0032] Figure 1 A schematic flow chart of the gearbox condition monitoring method provided in Example 1;

[0033] Figure 2 is a comparison curve diagram of the current energy ratio under different working conditions obtained by the monitoring method in Example 1;

[0034] Figure 3 This is a bar graph of the current-energy ratio obtained by applying the rotation speed estimation method in Example 2. DETAILED DESCRIPTION

[0035] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0036] The present invention provides a gearbox status monitoring method based on a stator current signal of a doubly-fed wind turbine to solve the problem. The method comprises:

[0037] Step S1: Read the current signal and speed signal of the wind turbine in the time domain, and sample the current data and speed data from the current signal and speed signal; in this embodiment, the sampling frequency of the current data is 10000Hz, and the sampling frequency of the speed data is 2Hz, and the data is collected continuously for 1 hour. The initial input data is random wind speed data of 8m / s-12m / s, which is taken from the actual wind field wind speed data and applied to the doubly fed wind power generation physical simulation platform. In general, the actual collected data is slightly different from the theoretical one and must be preprocessed into a unified format.

[0038] Step S2: Split the current data using zero-crossing detection, the number of current data splits being equal to the number of speed data, so that the subsequent reorganized data can be continuous in the time domain, thereby alleviating interference caused by spectrum energy leakage. This embodiment splits the current data to effectively solve the problem of frequency band aliasing under variable speed data.

[0039] Step S3: Matching the rotation speed data with the segmented current data; since the current data and the rotation speed data are acquired at different sampling frequencies, it is necessary to match the rotation speed data with the segmented current data on the time axis.

[0040] Step S4: Based on the speed data as a label; all current data matched by the speed data within the same preset speed range are divided into the same group; although the current data segmentation in this embodiment can avoid the frequency band aliasing problem of variable speed data, it also causes the data after segmentation to have insufficient spectrum accuracy after Fourier transform, which is caused by insufficient data volume, so it is necessary to reorganize multiple segmented data. The reorganization range is 51.5Hz-59.5Hz (corresponding to the speed range of 1545r / min-1785r / min), and each interval of 1Hz (30r / min) is a reorganization range, which can not only ensure that the reorganized signal does not have the frequency band aliasing problem, but also reduce the demand for data to a certain extent. The speed and frequency range can be pre-set according to the actual structure of the wind turbine, and the corresponding relationship between the two depends on the actual structure of the wind turbine. The corresponding relationship determined by the wind turbine structure is fixed, and the speed and frequency can be converted to each other through the determined corresponding relationship.

[0041] Step S5: Recombining a preset number of current data belonging to the same group to obtain recombined current data of each group; in a specific embodiment, the number of recombinations is defined as 30, that is, 30 data of the same group are recombined into one recombined data, which minimizes the demand for data while meeting the spectrum accuracy. Based on the sampling frequency and sampling time in step S1, theoretically, a total of 60*120 pieces of segmented data are required to ensure sufficient recombined data.

[0042] The state monitoring of the gearbox in this embodiment is based on the current energy ratio algorithm. The algorithm has a good monitoring effect under constant working conditions, but frequency band overlap will occur under variable working conditions and quantitative analysis cannot be performed. Steps S1-S5 have been able to solve the problem that the current energy ratio algorithm is difficult to perform under variable working conditions. The main principle of the current energy ratio algorithm is that if the performance of the gearbox degrades or fails, it will cause the torque on the main transmission chain to change, and eventually it will be input into the generator to cause the electromagnetic frequency and its double frequency band energy to change. The current energy ratio algorithm in this embodiment is inspired by the vibration signal sideband energy ratio algorithm, and in this embodiment, the electromagnetic frequency and double frequency are used to calculate the current energy ratio to achieve better monitoring effects in the double-fed grid-connected state.

[0043] Step S6: Based on the pre-established phenomenological model of the stator current signal of the doubly-fed wind turbine, the current energy ratio is calculated according to the reorganized current data of each group; in this embodiment, the current energy ratio is calculated by using the electromagnetic frequency double frequency to the electromagnetic frequency single frequency; the stator current signal modulation phenomenon is related to the speed of the doubly-fed machine, and the frequency of the pole logarithm multiple of the speed appears in the signal spectrum, which is defined as the electromagnetic frequency:

[0044] erf=p*f h

[0045] In the formula, p represents the number of pole pairs of the generator, f h It represents the high-speed shaft rotation frequency, and erf represents the electromagnetic rotation frequency.

[0046] There are differences in the amplitudes of all modulation phenomena of the stator current signal. It is proposed to use a frequency band with a larger amplitude to establish a phenomenological model. The formula of the model in step S6 is:

[0047] I=a1sin(2×50×π×t)+a2sin(2×erf×π×t)+a3sin(2×(2×(erf-50)+50)×π×t)+

[0048] a4sin(2×(6×(erf-50)+50)×π×t)+a5×sin(2×(6×(erf-50)+5×50)×π×t)+

[0049] a7sin(2×(6×(erf-50)+7×50)×π×t)+a 11 sin(2×(2×6×(erf-50)+11×50)×

[0050] π×t)+a 13 sin(2×(2×6×(erf-50)+13×50)×π×t)+a 17 sin(2×(3×6×(erf-50)+17×50)×π×t)

[0051] Where I represents the stator current signal, which is a superposition of multiple sine or cosine signals; a i represents the modulation amplitude, t represents time, erf represents electromagnetic frequency, and 50Hz represents power frequency;

[0052] The formula for the current energy ratio is:

[0053]

[0054] Wherein, Secondsidebandenergy represents the sideband energy corresponding to a3sin(2×(2×(erf-50)+50)×π×t) in the model, Firstsidebandenergy represents the sideband energy corresponding to a2sin(2×erf×π×t) in the model, and the sideband energy represents the sum of the spectrum amplitudes within the corresponding frequency range. The model formula in this embodiment is used to locate the sideband position, and after the sideband position is determined, to obtain the corresponding sideband energy (i.e., the sum of the spectrum amplitudes within the corresponding frequency range), only the sum function is required. Similarly, the frequency band range of the reorganized data is completely fixed, and to obtain the current energy ratio, it is only necessary to calculate the ratio based on the sum.

[0055] The model in this embodiment only considers the frequency modulation part of the model, and does not consider the amplitude modulation part of the model. The model can currently locate the frequency band with a higher amplitude within 1000Hz, and this model is applicable to both variable and constant operating conditions, normal and fault conditions. There are two types of modulation phenomena in this model. The first type is the electromagnetic frequency conversion and its double frequency band; the second type is the modulation phenomenon band caused by the motor tooth harmonics, the six times and its double frequency and the power frequency double frequency. This embodiment uses the electromagnetic frequency conversion double frequency to the electromagnetic frequency conversion single frequency to calculate the current energy ratio.

[0056] Step S7: Determine the state of the gearbox according to the mean of the current energy ratio obtained in each group. In this embodiment, the current energy ratio of multiple groups of reorganized data is calculated to obtain the mean as the final output result, which avoids the influence of singular values ​​to a certain extent; if a certain range is less than the preset number 30 and cannot be reorganized, it will not be calculated and a null value will be output. The current energy ratio index is theoretically generated in one group every hour. If the mean of the current energy ratio is greater than the pre-determined current energy ratio of the normal state, it is determined that there is a fault in the state of the gearbox. The larger the value of the current energy ratio, the higher the degree of deterioration of the gearbox and the higher the severity of the fault.

[0057] Example 2

[0058] The present embodiment is different from the present embodiment in that the speed data in step S1 is estimated by current data in the present embodiment. The speed estimation specifically includes the following steps:

[0059] Step 1: Divide the current data per minute into 60 parts corresponding to each second;

[0060] Step 2: Convert the current data per second into a current frequency domain signal through Fourier transform. The segmentation of the current data will lead to a decrease in the spectrum resolution. In this embodiment, the spectrum resolution is improved by adding zeros.

[0061] Step 3: Obtain the local maximum energy value of the high-frequency part of the current frequency domain signal; and obtain the rotation speed data per second based on the local maximum energy value.

[0062] According to the phenomenological model of the stator current signal in Example 1, the high-frequency part is the motor tooth harmonic, which has a certain corresponding relationship with the speed. The second-level speed of the wind turbine can be estimated based on the harmonic frequency. Figure 3 The figure shows the effect diagram of the current energy ratio algorithm obtained by the speed estimation method in this embodiment. Although the effect is not as good as the original data, the current energy ratio in the fault state is still stably higher than the current energy ratio in the normal state.

[0063] Through the description of the above implementation methods, those skilled in the art can clearly understand that the present invention can be implemented by hardware, or by means of software plus a necessary general hardware platform. Based on such an understanding, the technical solution of the present invention can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.), including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each implementation scenario of the present invention.

[0064] Those skilled in the art will appreciate that the modules in the device in the implementation scenario can be distributed in the device in the implementation scenario according to the implementation scenario description, or can be changed accordingly and located in one or more devices different from the implementation scenario. The modules in the above implementation scenario can be combined into one module, or can be further split into multiple sub-modules.

[0065] A computer-readable storage medium stores one or more programs, which, when executed by an electronic device including a plurality of application programs, cause the electronic device to execute the above method.

[0066] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit it. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A gearbox status monitoring method based on a doubly-fed wind turbine stator current signal, characterized in that: include: Step S1: reading a current signal and a speed signal of a wind turbine generator set in the time domain, and sampling the current signal and the speed signal to obtain current data and speed data; Step S2: dividing the current data by zero-crossing detection, wherein the number of the current data divided is equal to the number of the speed data; Step S3: matching the rotation speed data with the divided current data; Step S4: Based on the rotation speed data as a label, all current data matched by the rotation speed data belonging to the same preset rotation speed range are divided into the same group; Step S5: reorganizing a preset number of current data belonging to the same group to obtain reorganized current data of each group; Step S6: based on the pre-established phenomenological model of the stator current signal of the doubly-fed wind turbine generator set, respectively calculating the current energy ratio according to the reorganized current data of each group; Step S7: determining the state of the gearbox according to the average value of the current energy ratio obtained in each group; The formula of the model in step S6 is: I=a1sin(2×50×π×t)+a2sin(2×erf×π×t)+a3sin(2×(2×(erf-50)+50)×π×t)+a4sin(2×(6×(erf-50)+50)×π×t)+a5sin(2×(6×(erf-50)+5×50)×π×t)+a7sin(2×(6×(erf-50)+7×50)×π×t)+a 11 sin(2×(2×6×(erf-50)+11×50)×π×t)+a 13 sin(2×(2×6×(erf-50)+13×50)×π×t)+a 17 sin(2×(3×6×(erf-50)+17×50)×π×t) Where I represents the stator current signal, a i represents the modulation amplitude, t represents time, erf represents electromagnetic frequency, and 50Hz represents power frequency; The formula for the current energy ratio is: In the formula, SER represents the current energy ratio, Second sideband energy represents the sideband energy corresponding to a3sin(2×(2×(erf-50)+50)×π×t) in the model, and First sideband energy represents the sideband energy corresponding to a2sin(2×erf×π×t) in the model. The sideband energy represents the sum of the spectrum amplitudes in the corresponding frequency range.

2. The gearbox condition monitoring method according to claim 1, characterized in that: In step S1, the sampling frequency of the current data is 10000 Hz, the sampling frequency of the rotation speed data is 2 Hz, and the data are collected continuously for 1 hour.

3. The gearbox condition monitoring method according to claim 1, characterized in that: The speed range is 1545r / min-1785r / min, and each 30r / min is a preset speed range.

4. The gearbox condition monitoring method according to claim 1, characterized in that: The preset number in step S5 is 30.

5. The gearbox condition monitoring method according to claim 1, characterized in that: Step S7 includes: if the average value of the current-energy ratio is greater than the pre-determined current-energy ratio of the normal state, determining that there is a fault in the state of the gearbox.

6. The gearbox condition monitoring method according to claim 1, characterized in that: The method for acquiring the speed data in step S1 is: Divide the current data per minute into 60 parts corresponding to each second; Convert the current data per second into a current frequency domain signal through Fourier transform, and obtain the local maximum energy value of the high frequency part of the current frequency domain signal; The rotation speed data per second is acquired according to the local maximum value of energy.

Images