CN117703688A - Wind driven generator variable working condition state monitoring method based on amplitude standardization statistical analysis - Google Patents

Abstract

A wind driven generator variable working condition state monitoring method based on amplitude standardization statistical analysis comprises the following steps: 1) Collecting vibration data in the running process of the wind driven generator in real time; 2) Calculating a statistical feature ratio coefficient; 3) Processing the vibration data to obtain vibration data with standardized amplitude; 4) Establishing a probability distribution model and determining corresponding parameters; 5) Inputting vibration data with standardized amplitude values into a probability distribution model to obtain probability distribution of the vibration data in the running process of the wind driven generator; 6) Comparing and analyzing the vibration data probability distribution of the wind driven generator in the running process with the reference probability distribution of the normal running stage, and calculating the probability distribution fluctuation value; 7) Judging whether the probability distribution fluctuation value exceeds a preset threshold range, if so, the current wind driven generator operates abnormally. The invention is beneficial to improving the operation and maintenance efficiency of the wind driven generator and the wind energy productivity, and has very urgent actual application requirements of the wind power industry and wide application prospects of the wind power industry.

Description

Wind driven generator variable working condition state monitoring method based on amplitude standardization statistical analysis

Technical Field

The invention relates to the field of health monitoring and intelligent operation and maintenance of mechanical equipment, in particular to a wind driven generator variable-working-condition state monitoring method based on amplitude standardized statistical analysis.

Background

As a new renewable energy source, wind energy not only optimizes the energy supply structure, reduces environmental pollution, but also reduces global carbon emissions, and has been increasingly focused and utilized in recent years. As an important device for wind energy conversion, the installed amount of wind power generators is also increasing year by year, and the wind power market is rapidly developing. However, it is noted that the problem of high operation and maintenance costs is also becoming more pronounced in order to maintain the normal continuous operation of the wind turbine.

In order to improve the service reliability of the wind driven generator and reduce the unexpected downtime, the operation and maintenance cost is further reduced, and the effective monitoring of the operation state of the wind driven generator has an extremely important role. However, the operation condition of the wind driven generator is very complex, and the variable condition causes difficulty in directly analyzing whether the operation of the wind driven generator is normal or not from the collected operation process data.

In view of this problem, expert scholars have conducted research from different angles in recent years, and the differences established according to the monitoring mechanism can be divided into two main categories: firstly, an operation state monitoring mechanism based on signal processing; and secondly, an operation state monitoring mechanism based on deep learning. The former reduces coupling influence caused by variable working conditions by carrying out operations such as envelope demodulation, sample data cutting and the like on collected operation process data, and an abnormal state early warning mechanism is arranged on the basis of the coupling influence, however, the method depends on setting key parameters such as sample data cutting length and the like, and the inhibition of variable working condition interference needs to be further improved. The method is mainly developed under a plurality of discrete working conditions, and is remarkable in that the running working conditions of the wind driven generator are complex and continuous, and the accuracy of the running state monitoring network obtained based on the method in the practical application process is required to be further improved.

The monitoring of the running state of the wind driven generator under the variable working condition is a troublesome problem, because the variable working condition can lead to distortion of the characteristic space, so that the traditional characteristic extraction means based on the characteristic space measurement is difficult to work, and further great difficulty is brought to the accurate representation and effective monitoring of the running state of the wind driven generator.

Disclosure of Invention

The invention aims to provide a wind driven generator variable working condition state monitoring method based on amplitude standardization statistical analysis, which comprises the following steps:

1) Vibration data in the running process of the wind driven generator are collected in real time.

2) And calculating a statistical feature ratio coefficient according to vibration data acquired in real time.

3) And carrying out amplitude standardization processing on vibration data acquired in real time according to the statistical characteristic ratio coefficient to obtain a vibration data sample with standardized amplitude.

4) And establishing a probability distribution model for describing the vibration data statistical characteristics of the wind driven generator, and determining parameters in the probability distribution model.

5) And inputting the vibration data sample with standardized amplitude into a probability distribution model to obtain the probability distribution of the vibration data in the running process of the wind driven generator.

6) And comparing and analyzing the vibration data probability distribution of the running process of the wind driven generator obtained at the current moment with the reference probability distribution of the normal running stage, and calculating the probability distribution fluctuation value of the vibration data sample with the standardized amplitude at the current moment.

7) Judging whether the fluctuation value exceeds a preset threshold range, if so, judging that the current wind driven generator operates abnormally, and if not, the wind driven generator operates normally.

Further, vibration data in the running process of the wind driven generator is collected through a sensor arranged on the wind driven generator.

Further, the step of calculating the statistical feature ratio coefficient according to the vibration data acquired in real time includes:

2.1 Calculating a reference sample x _b (n) as follows:

wherein i is the acquisition period sequence number, x _i-1 (n) is wind driven generator operation process vibration data acquired in the (i-1) th acquisition period. X is x _r (n) vibration of the history wind driven generator in the normal operation stageA dynamic data sample. n is the sample length number of the vibration data. L represents the total length of the sample of vibration data. t is t _i Representing the time of the current acquisition period. t is t _M Representing the maintenance time of the wind power generator.

2.2 Calculating a statistical feature ratio coefficient as follows:

in the method, in the process of the invention,is a statistical feature ratio coefficient. X is x _i (n) vibration data of the wind driven generator operation process acquired in the ith acquisition period. Omega _i (t) represents rotational speed information of the i-th acquisition period. Omega _b (t) rotational speed information representing a reference sample acquisition period. τ (·) represents the statistical signature function. K represents the total number of samples which are close to the reference sample and have the statistical characteristic values exceeding the statistical characteristic of vibration data in the ith acquisition period; b is a reference sample number; k is the vibration data sample number adjacent to the reference sample.

Further, the amplitude normalized vibration data samples are shown below:

where n is the sample length number of the vibration data. Phi [ x ] _i (n),x _b (n),t _M ,K]Is a statistical feature ratio coefficient. X is x _i (n) vibration data of the wind driven generator operation process acquired in the ith acquisition period.Representative of a vibration data sample of normalized amplitude.

Further, the probability distribution model for describing the vibration data statistical characteristics of the wind driven generator is as follows:

wherein phi (t) is the probability distribution of vibration data in the running process of the wind driven generator. Alpha is a characteristic index, beta is a symmetrical parameter, gamma is a scale parameter, delta is a position parameter, and j is an imaginary unit.

t is the regression coefficient.

Wherein the sign function sgn (t) is as follows:

further, the regression coefficients t are as follows:

where Λ (·) represents the interpolation function and ζ is the step size. n is the sample length number of the vibration data.Representative of a vibration data sample of normalized amplitude.

Further, the step of determining parameters in the probability distribution model includes:

4.1 Establishment of maximum log likelihood functionThe following is shown:

where n is the sample length number of the vibration data. L represents the total length of the sample of vibration data.Representative amplitude normalization Probability density function of vibration data samples, +.>Representative of a vibration data sample of normalized amplitude. Θ is a model parameter, Θ= { α, β, γ, δ }, α is a feature index, β is a symmetry parameter, γ is a scale parameter, and δ is a position parameter.

4.2 Calculating a probability density function for sample data centralization by a numerical integration methodThe following is shown:

wherein the sample data z is centralized _n The following is shown:

the parameter v (θ) is as follows:

wherein, θ is a variable, and the value range is [0, pi/2 ].

4.3 Estimating the model parameters Θ by means of a maximum log-likelihood function, as follows:

further, the step of comparing and analyzing the probability distribution of vibration data of the wind driven generator operation process obtained at the current moment with the reference probability distribution of the normal operation stage, and calculating the fluctuation value of the probability distribution at the current moment comprises the following steps:

6.1 Determining the boundaries of the probability distribution model as follows:

in the method, in the process of the invention,a vibration data sample representative of the amplitude normalization of the ith acquisition period. />Vibration data samples representing the normalized amplitude of the normal operating phase of the historical wind turbine. Theta (theta) _i Representing vibration data +.>Is used for the model parameters of the model. Theta (theta) ₀ Representative vibration data sample->Is used for the model parameters of the model. l (L) _R 、l _L The upper and lower boundaries of the probability distribution model, respectively.

Wherein the correction term Δμ is as follows:

6.2 Calculating the probability distribution fluctuation value of the vibration data sample with the normalized amplitude at the current moment, as follows:

in the method, in the process of the invention,the probability distribution fluctuation value of the vibration data sample normalized for the amplitude of the current moment.

Further, the setting step of the threshold range includes:

7.1 Continuously collecting N vibration data samples of the normal operation stage of the wind driven generator, and respectively calculating corresponding probability distribution fluctuation values.

7.2 A mean value and a standard deviation of probability distribution fluctuation values of the N vibration data samples are calculated.

7.3 A 3 sigma threshold interval is set according to the mean value and standard deviation of probability distribution fluctuation values of the N vibration data samples.

Further, if the current wind driven generator is judged to be abnormal in operation, a spectrum analysis method is adopted for verification.

If the fault characteristic frequency is obvious, the wind driven generator fails.

The method for monitoring the running state of the wind driven generator has the advantages that the method is applicable to variable working conditions and high in accuracy, the running state monitoring method is beneficial to further improving the running and maintenance efficiency of the wind driven generator, the wind energy productivity is improved, and the method has very urgent practical application requirements of the wind power industry and very wide application prospects of the wind power industry.

The beneficial effects of the invention are as follows:

aiming at the variable working condition operation process of the wind driven generator, the invention provides a variable working condition state monitoring method of the wind driven generator based on amplitude standardization statistical analysis, and an amplitude standardization strategy is established by determining a reference sample and calculating a statistical characteristic ratio coefficient, so that the effective demodulation of amplitude fluctuation caused by variable working conditions is realized; in addition, a probability distribution model for describing the vibration data statistics characteristics of the wind driven generator is established, and a maximum log likelihood estimation method is adopted to determine characteristic model parameters; on the basis, comparing and analyzing the vibration data probability distribution of the running process of the wind driven generator obtained at the current moment with the reference probability distribution of the normal running stage, calculating the fluctuation value of the probability distribution at the current moment, and judging whether the running state of the wind driven generator is normal or not according to the fluctuation value. The method provided by the invention can be effectively applied to a wind driven generator variable working condition state monitoring scene, overcomes the defect that the traditional method is limited by a constant working condition factor, and effectively reduces state false alarm caused by working condition fluctuation. The invention has remarkable technical effect, good engineering should value and wide application prospect.

Drawings

FIG. 1 is a logic diagram of a wind turbine variable condition state monitoring method based on amplitude standardization statistical analysis;

FIG. 2 is a time domain waveform of vibration data of a bearing at a driving end of a wind driven generator actually collected in an embodiment of the invention;

FIG. 3 is a diagram showing the corresponding information of the rotational speed of the main shaft of the generator during the process of collecting vibration data of the bearing at the driving end of the wind driven generator according to the embodiment of the invention;

FIG. 4 is a graph showing the monitoring result of the bearing variable working condition state of the driving end of the wind driven generator based on the amplitude standardized statistical analysis in the embodiment of the invention;

FIG. 5 is a graph showing the results of a spectral analysis of vibration data of a bearing at the drive end of a wind turbine in accordance with an embodiment of the present invention;

FIG. 6 is a graph showing the results of monitoring the condition of a bearing at the drive end of a wind turbine according to various methods of the present invention; FIG. 6 (a) is a root mean square view of the condition monitoring of the bearing of the drive end of the wind turbine; FIG. 6 (b) is a graph showing the result of monitoring the state of the bearing at the driving end of the wind turbine; FIG. 6 (c) is a graph showing the mobility versus condition monitoring of the bearing at the drive end of the wind turbine; fig. 6 (d) is a result of monitoring a state of an Average info gram on a bearing at a driving end of a wind turbine; FIG. 6 (e) is a graph showing the state monitoring result of Zhan Senrui vergence on the bearing of the driving end of the wind driven generator based on multi-scale fuzzy entropy; FIG. 6 (f) is a graph showing the monitoring result of the state of the bearing at the driving end of the wind turbine by the maximum value of the spectrum; FIG. 6 (g) is a graph showing the result of monitoring the state of the bearing at the drive end of the wind turbine by the spectrum variance; FIG. 6 (h) is a generalized expansion Hjorth parameter monitoring result of the state of the bearing at the driving end of the wind driven generator; FIG. 6 (i) is a graph showing the monitoring result of the bearing condition of the driving end of the wind turbine according to the present invention;

FIG. 7 is a graph comparing ROC curves of the monitoring results of the driving end bearing state of the wind driven generator according to various methods of the embodiment of the invention.

Detailed Description

The present invention is further described below with reference to examples, but it should not be construed that the scope of the above subject matter of the present invention is limited to the following examples. Various substitutions and alterations are made according to the ordinary skill and familiar means of the art without departing from the technical spirit of the invention, and all such substitutions and alterations are intended to be included in the scope of the invention.

Example 1:

referring to fig. 1 to 7, a wind driven generator variable working condition state monitoring method based on amplitude standardization statistical analysis comprises the following steps:

1) Vibration data in the running process of the wind driven generator are collected in real time.

2) And calculating a statistical feature ratio coefficient according to vibration data acquired in real time.

3) And carrying out amplitude standardization processing on vibration data acquired in real time according to the statistical characteristic ratio coefficient to obtain a vibration data sample with standardized amplitude.

4) And establishing a probability distribution model for describing the vibration data statistical characteristics of the wind driven generator, and determining parameters in the probability distribution model.

5) And inputting the vibration data sample with standardized amplitude into a probability distribution model to obtain the probability distribution of the vibration data in the running process of the wind driven generator.

6) And comparing and analyzing the vibration data probability distribution of the running process of the wind driven generator obtained at the current moment with the reference probability distribution of the normal running stage, and calculating the probability distribution fluctuation value of the vibration data sample with the standardized amplitude at the current moment.

7) Judging whether the fluctuation value exceeds a preset threshold range, if so, judging that the current wind driven generator operates abnormally, and if not, the wind driven generator operates normally.

Example 2:

a wind driven generator variable working condition state monitoring method based on amplitude standardization statistical analysis is disclosed in the embodiment 1, and further, vibration data in the running process of the wind driven generator is collected through a sensor arranged on the wind driven generator.

Example 3:

the method for monitoring the variable working condition state of the wind driven generator based on the amplitude standardization statistical analysis has the main technical content as shown in any one of the embodiments 1 to 2, and further, the step of calculating the statistical characteristic ratio coefficient according to the vibration data collected in real time comprises the following steps:

2.1 Calculating a reference sample x _b (n) as follows:

wherein i is the acquisition period sequence number, x _i-1 (n) is wind driven generator operation process vibration data acquired in the (i-1) th acquisition period. X is x _r And (n) is a vibration data sample of the normal operation stage of the historical wind driven generator. n is the sample length number of the vibration data. L represents the total length of the sample of vibration data. t is t _i Representing the time of the current acquisition period. t is t _M Representing the maintenance time of the wind power generator.

2.2 Calculating a statistical feature ratio coefficient as follows:

in the method, in the process of the invention,is a statistical feature ratio coefficient. X is x _i (n) vibration data of the wind driven generator operation process acquired in the ith acquisition period. Omega _i (t) represents rotational speed information of the i-th acquisition period. Omega _b (t) rotational speed information representing a reference sample acquisition period. />Representing a statistical characteristic function. K represents the clinical value of the reference sampleThe number of the near statistical features exceeds the total number of samples of the vibration data statistical features of the ith acquisition period; b is a reference sample number; k is the vibration data sample number adjacent to the reference sample.

Example 4:

the main technical content of the wind driven generator variable working condition state monitoring method based on amplitude standardization statistical analysis is as shown in any one of embodiments 1 to 3, and further, vibration data samples with standardized amplitude are as follows:

where n is the sample length number of the vibration data. Phi [ x ] _i (n),x _b (n),t _M ,K]Is a statistical feature ratio coefficient. X is x _i (n) vibration data of the wind driven generator operation process acquired in the ith acquisition period.Representative of a vibration data sample of normalized amplitude.

Example 5:

the main technical content of the wind driven generator variable working condition state monitoring method based on the amplitude standardization statistical analysis is as shown in any one of embodiments 1 to 4, and further, the probability distribution model for describing the vibration data statistical characteristics of the wind driven generator is as follows:

where φ (t) is the probability of vibration data. Alpha is a characteristic index, beta is a symmetrical parameter, gamma is a scale parameter, delta is a position parameter, and j is an imaginary unit. t is the regression coefficient.

Wherein the sign function sgn (t) is as follows:

example 6:

the main technical content of the wind driven generator variable working condition state monitoring method based on the amplitude standardization statistical analysis is as shown in any one of embodiments 1 to 5, and further, the regression coefficient t is as follows:

where Λ (·) represents the interpolation function and ζ is the step size. n is the sample length number of the vibration data.Representative of a vibration data sample of normalized amplitude.

Example 7:

the method for monitoring the variable working condition state of the wind driven generator based on the amplitude standardization statistical analysis has the main technical content as shown in any one of embodiments 1 to 6, and further, the step of determining the parameters in the probability distribution model comprises the following steps:

4.1 Establishment of maximum log likelihood functionThe following is shown:

where n is the sample length number of the vibration data. L represents the total length of the sample of vibration data.Probability density function representing amplitude normalized vibration data samples, +.>Representative of a vibration data sample of normalized amplitude. Θ is a model parameter, Θ= { α, β, γ, δ }, α is a feature index,beta is a symmetrical parameter, gamma is a scale parameter, and delta is a position parameter.

4.2 Calculating a probability density function for sample data centralization by a numerical integration methodThe following is shown:

wherein the sample data z is centralized _n The following is shown:

the parameter v (θ) is as follows:

wherein, θ is a variable, and the value range is [0, pi/2 ].

4.3 Estimating the model parameters Θ by means of a maximum log-likelihood function, as follows:

example 8:

the method for monitoring the variable working condition state of the wind driven generator based on the amplitude standardization statistical analysis has the main technical content as shown in any one of the embodiments 1 to 7, and further, the step of comparing and analyzing the probability distribution of vibration data of the wind driven generator operation process obtained at the current moment with the reference probability distribution of the normal operation stage, and calculating the fluctuation value of the probability distribution at the current moment comprises the following steps:

6.1 Determining the boundaries of the probability distribution model as follows:

in the method, in the process of the invention,a vibration data sample representative of the amplitude normalization of the ith acquisition period. />Vibration data samples representing the normalized amplitude of the normal operating phase of the historical wind turbine. Theta (theta) _i Representing vibration data +.>Is used for the model parameters of the model. Theta (theta) ₀ Representative vibration data sample->Is used for the model parameters of the model. l (L) _R 、l _L The upper and lower boundaries of the probability distribution model, respectively.

Wherein the correction term Δμ is as follows:

6.2 Calculating the probability distribution fluctuation value of the vibration data sample with the normalized amplitude at the current moment, as follows:

in the method, in the process of the invention,probability score for vibration data samples normalized for current time amplitudeCloth waving value.

Example 9:

the method for monitoring the variable working condition state of the wind driven generator based on the amplitude standardization statistical analysis mainly comprises the following steps of:

7.1 Continuously collecting N vibration data samples of the normal operation stage of the wind driven generator, and respectively calculating corresponding probability distribution fluctuation values.

7.2 A mean value and a standard deviation of probability distribution fluctuation values of the N vibration data samples are calculated.

7.3 A 3 sigma threshold interval is set according to the mean value and standard deviation of probability distribution fluctuation values of the N vibration data samples.

Example 10:

the method for monitoring the variable working condition state of the wind driven generator based on the amplitude standardization statistical analysis mainly comprises the following steps of any one of the embodiments 1 to 9, and further, if the current wind driven generator is judged to be abnormal in operation, a spectrum analysis method is adopted for verification.

If the fault characteristic frequency is obvious, the wind driven generator fails.

Example 11:

referring to fig. 1 to 7, a wind driven generator variable working condition state monitoring method based on amplitude standardization statistical analysis comprises the following steps:

1) Vibration data in the running process of the wind driven generator are collected in real time.

2) And calculating a statistical feature ratio coefficient according to vibration data acquired in real time.

3) And carrying out amplitude standardization processing on vibration data acquired in real time according to the statistical characteristic ratio coefficient to obtain a vibration data sample with standardized amplitude.

4) And establishing a probability distribution model for describing the vibration data statistical characteristics of the wind driven generator, and determining parameters in the probability distribution model.

5) And inputting the vibration data sample with standardized amplitude into a probability distribution model to obtain the probability distribution of the vibration data in the running process of the wind driven generator.

6) And comparing and analyzing the vibration data probability distribution of the running process of the wind driven generator obtained at the current moment with the reference probability distribution of the normal running stage, and calculating the probability distribution fluctuation value of the vibration data sample with the standardized amplitude at the current moment.

7) Judging whether the fluctuation value exceeds a preset threshold range, if so, judging that the current wind driven generator operates abnormally, and if not, the wind driven generator operates normally.

Example 12:

a wind driven generator variable working condition state monitoring method based on amplitude standardization statistical analysis is disclosed in the main technical content of the embodiment 11, and further, vibration data in the running process of the wind driven generator is collected through a sensor arranged on the wind driven generator.

Example 13:

the method for monitoring the variable working condition state of the wind driven generator based on the amplitude standardization statistical analysis has the main technical content as shown in any one of embodiments 11 to 12, and further, the step of calculating the statistical characteristic ratio coefficient according to vibration data collected in real time comprises the following steps:

2.1 Calculating a reference sample x _b (n) as follows:

wherein i is the acquisition period sequence number, x _i-1 (n) is wind driven generator operation process vibration data acquired in the (i-1) th acquisition period. X is x _r And (n) is a vibration data sample of the normal operation stage of the historical wind driven generator. n is the sample length number of the vibration data. L represents the total length of the sample of vibration data. t is t _i Representing the time of the current acquisition period. t is t _M Representing the maintenance time of the wind power generator.

2.2 Calculating a statistical feature ratio coefficient as follows:

in the method, in the process of the invention,represents a statistical characteristic function omega _i (t) and ω _b (t) rotational speed information representing the current acquisition period and the reference sample acquisition period, respectively.

3) In order to eliminate the influence of abnormal interference and improve the robustness of amplitude normalization, the statistical feature ratio coefficient is further set as:

in the method, in the process of the invention,is a statistical feature ratio coefficient. X is x _i (n) vibration data of the wind driven generator operation process acquired in the ith acquisition period. Omega _i (t) represents rotational speed information of the i-th acquisition period. Omega _b (t) rotational speed information representing a reference sample acquisition period. />Representing a statistical characteristic function. K represents the total number of samples which are close to the reference sample and have the statistical characteristic values exceeding the statistical characteristic of vibration data in the ith acquisition period; b is a reference sample number; k is the vibration data sample number adjacent to the reference sample.

Example 14:

the main technical content of the wind driven generator variable working condition state monitoring method based on the amplitude standardization statistical analysis is as shown in any one of embodiments 11 to 13, and further, vibration data samples with standardized amplitude are as follows:

wherein n is the sample length number of the vibration data 。Φ[x _i (n),x _b (n),t _M ,K]Is a statistical feature ratio coefficient. X is x _i (n) vibration data of the wind driven generator operation process acquired in the ith acquisition period.Representative of a vibration data sample of normalized amplitude.

Example 15:

the main technical content of the wind driven generator variable working condition state monitoring method based on the amplitude standardization statistical analysis is as shown in any one of embodiments 11 to 14, and further, the probability distribution model for describing the vibration data statistical characteristics of the wind driven generator is as follows:

where φ (t) is the probability of vibration data. Alpha is a characteristic index, beta is a symmetrical parameter, gamma is a scale parameter, delta is a position parameter, and j is an imaginary unit. t is the regression coefficient.

Wherein the sign function sgn (t) is as follows:

example 16:

the main technical content of the wind driven generator variable working condition state monitoring method based on the amplitude standardization statistical analysis is as shown in any one of embodiments 11 to 15, and further, the regression coefficient t is obtained through an interpolation method, and the method is as follows:

where Λ (·) represents the interpolation function and ζ is the step size. n is the sample length number of the vibration data.Representative of a vibration data sample of normalized amplitude.

Example 17:

the method for monitoring the variable working condition state of the wind driven generator based on the amplitude standardization statistical analysis has the main technical content as shown in any one of embodiments 11 to 16, and further, the step of determining parameters in the probability distribution model comprises the following steps:

4.1 Establishment of maximum log likelihood functionThe following is shown:

where n is the sample length number of the vibration data. L represents the total length of the sample of vibration data.Probability density function representing amplitude normalized vibration data samples, +.>Representative of a vibration data sample of normalized amplitude. Θ is a model parameter, Θ= { α, β, γ, δ }, α is a feature index, β is a symmetry parameter, γ is a scale parameter, and δ is a position parameter. />

4.2 Calculating a probability density function for sample data centralization by a numerical integration methodThe following is shown:

wherein the sample data z is centralized _n The following is shown:

the parameter v (θ) is as follows:

wherein, θ is a variable, and the value range is [0, pi/2 ].

4.3 Estimating the model parameters Θ by means of a maximum log-likelihood function, as follows:

example 18:

the method for monitoring the variable working condition state of the wind driven generator based on the amplitude standardization statistical analysis has the main technical content as shown in any one of embodiments 11 to 17, and further, the step of comparing and analyzing the probability distribution of vibration data of the wind driven generator operation process obtained at the current moment with the reference probability distribution of the normal operation stage, and calculating the fluctuation value of the probability distribution at the current moment comprises the following steps:

6.1 Determining the boundaries of the probability distribution model as follows:

in the method, in the process of the invention,a vibration data sample representative of the amplitude normalization of the ith acquisition period. />Representing historyVibration data samples with standardized amplitude values at the normal operation stage of the wind driven generator. Theta (theta) _i Representing vibration data +.>Is used for the model parameters of the model. Theta (theta) ₀ Representative vibration data sample->Is used for the model parameters of the model. l (L) _R 、l _L The upper and lower boundaries of the probability distribution model, respectively.

Wherein the correction term Δμ is as follows:

6.2 Calculating the probability distribution fluctuation value of the vibration data sample with the normalized amplitude at the current moment, as follows:

in the method, in the process of the invention,the probability distribution fluctuation value of the vibration data sample normalized for the amplitude of the current moment.

Example 19:

the method for monitoring the variable working condition state of the wind driven generator based on the amplitude standardization statistical analysis mainly comprises the following steps of:

7.1 Continuously collecting N vibration data samples of the normal operation stage of the wind driven generator, and respectively calculating corresponding probability distribution fluctuation values.

7.2 A mean value and a standard deviation of probability distribution fluctuation values of the N vibration data samples are calculated.

7.3 A 3 sigma threshold interval is set according to the mean value and standard deviation of probability distribution fluctuation values of the N vibration data samples.

Calculating probability distribution fluctuation values of vibration data of the wind driven generator in the running process of the subsequent continuous collection by using a formula (16), and if the probability distribution fluctuation values do not exceed a 3 sigma threshold interval, indicating that the running state of the wind driven generator is normal; otherwise, the running process is abnormal.

Example 20:

the main technical content of the wind driven generator variable working condition state monitoring method based on the amplitude standardization statistical analysis is as shown in any one of embodiments 11 to 19, and further, if the current wind driven generator is judged to be abnormal in operation, a spectrum analysis method is adopted for verification.

If the fault characteristic frequency is obvious, the wind driven generator fails.

Example 21:

referring to fig. 1 to 7, a wind driven generator variable working condition state monitoring method based on amplitude standardization statistical analysis comprises the following steps:

1) Vibration data in the running process are collected by using a sensor arranged on the wind driven generator.

2) And establishing an amplitude standardization strategy by determining a reference sample and calculating a statistical characteristic ratio coefficient according to vibration data of the running process of the wind driven generator acquired at the current moment.

3) And carrying out amplitude standardization on vibration data of the running process of the wind driven generator, which is acquired at the current moment, according to the established amplitude standardization strategy.

4) And establishing a probability distribution model for describing the vibration data statistical characteristics of the wind driven generator, and determining parameters in the characteristic model by adopting a maximum log likelihood estimation method.

5) Determining a probability distribution statistical analysis boundary, comparing and analyzing the probability distribution of vibration data of the wind driven generator operation process obtained at the current moment with the reference probability distribution of the normal operation stage, and calculating the fluctuation value of the probability distribution at the current moment.

6) And determining a threshold range of the probability distribution fluctuation value, and judging the running state of the wind driven generator at the current moment according to the calculated probability distribution fluctuation value.

Aiming at vibration data of the running process of the wind driven generator acquired at the current moment, the step of establishing an amplitude standardization strategy by determining a reference sample and calculating a statistical characteristic ratio coefficient comprises the following steps:

1) Vibration data x of wind driven generator operation process acquired at current moment (set as ith acquisition period) _i Calculate a reference sample x _b ：

Wherein L represents the sample length of vibration data, x _i-1 Vibration data, x, of the running process of the wind driven generator, which are acquired in the (i-1) th acquisition period _r For vibration data samples of a preselected normal operation phase of the wind turbine, t _i And t _M Representing the time of the current acquisition period and the maintenance time of the wind turbine, respectively.

2) Calculating a statistical feature ratio coefficient:

in the method, in the process of the invention,represents a statistical characteristic function omega _i (t) and ω _b (t) rotational speed information representing the current acquisition period and the reference sample acquisition period, respectively.

3) In order to eliminate the influence of abnormal interference and improve the robustness of amplitude normalization, the statistical feature ratio coefficient is further set as:

wherein K represents the total number of samples adjacent to the reference sample and having a statistical characteristic value exceeding the statistical characteristic of vibration data of the ith acquisition period. b is the determined reference sample number. k is the vibration data sample number adjacent to the reference sample.

The method for carrying out amplitude standardization on vibration data of the running process of the wind driven generator acquired at the current moment according to the established amplitude standardization strategy comprises the following steps:

in the method, in the process of the invention,representing wind turbine operational process vibration data after the amplitude normalization operation.

The step of establishing a probability distribution model for describing the vibration data statistical characteristics of the wind driven generator and determining parameters in the feature model by adopting a maximum log likelihood estimation method comprises the following steps of:

1) Establishing a probability distribution model for describing the vibration data statistical characteristics of the wind driven generator:

Wherein, alpha is a characteristic index, beta is a symmetrical parameter, gamma is a scale parameter, delta is a position parameter, and j is an imaginary unit. sgn (t) is a sign function, and the calculation formula is:

in addition, the parameter t represents a regression coefficient, which can be obtained by an interpolation method:

where Λ (·) represents the interpolation function and ζ is the step size.

2) Establishing a maximum log-likelihood function:

in the method, in the process of the invention,and a probability density function representing vibration data of the wind driven generator running process after the amplitude normalization operation, wherein Θ= { alpha, beta, gamma, delta } is a model parameter.

3) Calculation by numerical integration method

Wherein z is _n The representative sample data centralization operation is as follows:

in addition, the calculation method of v (θ) in the formula (9) is:

wherein, the value range of theta is [0, pi/2 ].

4) Estimating the model parameters Θ= { α, β, γ, δ }, by maximizing the log-likelihood function:

determining a probability distribution statistical analysis boundary, comparing and analyzing the probability distribution of vibration data of the wind driven generator operation process obtained at the current moment with the reference probability distribution of a normal operation stage, and calculating the fluctuation value of the probability distribution at the current moment, wherein the step of calculating the fluctuation value of the probability distribution at the current moment comprises the following steps:

1) Determining a probability distribution statistical analysis boundary:

/>

in the method, in the process of the invention,representative of a vibration data sample obtained during a current data acquisition period and after an amplitude normalization operation; theta (theta) _i Representative vibration data sample->Model parameters of (2); />Representing vibration data samples collected in a normal operation stage of the historical wind driven generator and subjected to amplitude standardization operation; theta (theta) ₀ Representative vibration data sample->Is used for the model parameters of the model. l (L) _R 、l _L The upper and lower boundaries of the probability distribution model, respectively. Δμ is a correction term, and the calculation method is as follows:

2) Calculating the fluctuation value of the probability distribution at the current moment:

the method for determining the threshold range of the probability distribution fluctuation value comprises the following steps of:

1) N vibration data samples of the normal operation stage of the wind driven generator are continuously collected, and corresponding probability distribution fluctuation values are calculated respectively.

2) The mean and standard deviation of the N probability distribution fluctuation values are calculated, respectively, thereby setting a 3σ threshold interval.

3) Calculating probability distribution fluctuation values of vibration data of the wind driven generator in the running process of the subsequent continuous collection by using a formula (16), and if the probability distribution fluctuation values do not exceed a 3 sigma threshold interval, indicating that the running state of the wind driven generator is normal; otherwise, the running process is abnormal.

And verifying vibration data of the wind driven generator in the running process of the wind driven generator judged to be in an abnormal state by adopting a spectrum analysis method, and if the fault characteristic frequency is obvious, indicating that the wind driven generator has faults.

Example 22:

referring to fig. 1 to 7, a wind driven generator variable working condition state monitoring method based on amplitude standardization statistical analysis comprises the following steps:

1) Vibration data in the running process are collected by using a sensor arranged on the wind driven generator.

2) And establishing an amplitude standardization strategy by determining a reference sample and calculating a statistical characteristic ratio coefficient according to vibration data of the running process of the wind driven generator acquired at the current moment.

The step of establishing an amplitude normalization strategy by determining a reference sample and calculating a statistical feature ratio coefficient comprises:

2.1 Wind turbine generator operation process vibration data x acquired for the current time (set as the ith acquisition period) _i Calculate a reference sample x _b ：

Wherein L represents the sample length of vibration data, x _i-1 Vibration data, x, of the running process of the wind driven generator, which are acquired in the (i-1) th acquisition period _r For vibration data samples of a preselected normal operation phase of the wind turbine, t _i And t _M Representing the time of the current acquisition period and the maintenance time of the wind turbine, respectively.

2.2 Calculating a statistical feature ratio coefficient:

in the method, in the process of the invention,represents a statistical characteristic function omega _i (t) and ω _b (t) rotational speed information representing the current acquisition period and the reference sample acquisition period, respectively.

2.3 To eliminate the effect of abnormal interference, to improve the robustness of the amplitude normalization, the statistical feature ratio coefficient is further set to:

wherein K represents the total number of samples adjacent to the reference sample and having a statistical characteristic value exceeding the statistical characteristic of vibration data of the ith acquisition period. b is the determined reference sample number. k is the vibration data sample number adjacent to the reference sample.

3) And carrying out amplitude standardization on vibration data of the running process of the wind driven generator, which is acquired at the current moment, according to the established amplitude standardization strategy.

The method for carrying out amplitude standardization on vibration data of the running process of the wind driven generator acquired at the current moment according to the established amplitude standardization strategy comprises the following steps:

in the method, in the process of the invention,representing wind turbine operational process vibration data after the amplitude normalization operation.

4) And establishing a probability distribution model for describing the vibration data statistical characteristics of the wind driven generator, and determining parameters in the characteristic model by adopting a maximum log likelihood estimation method.

The method for establishing the probability distribution model for describing the vibration data statistical characteristics of the wind driven generator and determining the characteristic model parameters by adopting the maximum log likelihood estimation method comprises the following steps of:

4.1 Building a probability distribution model for describing the vibration data statistics of the wind driven generator:

wherein, alpha is a characteristic index, beta is a symmetrical parameter, gamma is a scale parameter, delta is a position parameter, and j is an imaginary unit. sgn (t) is a sign function, and the calculation formula is:

/>

in addition, the parameter t represents a regression coefficient, which can be obtained by an interpolation method:

where Λ (·) represents the interpolation function and ζ is the step size.

4.2 -establishing a maximum log likelihood function:

in the method, in the process of the invention,and a probability density function representing vibration data of the wind driven generator running process after the amplitude normalization operation, wherein Θ= { alpha, beta, gamma, delta } is a model parameter.

4.3 Calculating by numerical integration method

Wherein z is _n The representative sample data centralization operation is as follows:

in addition, the calculation method of v (θ) in the formula (9) is:

wherein, the value range of theta is [0, pi/2 ].

4.4 Estimating the model parameters Θ= { α, β, γ, δ }) by maximizing the log-likelihood function:

5) Determining a probability distribution statistical analysis boundary, comparing and analyzing the probability distribution of vibration data of the wind driven generator operation process obtained at the current moment with the reference probability distribution of the normal operation stage, and calculating the fluctuation value of the probability distribution at the current moment.

The step of determining the probability distribution statistical analysis boundary and calculating the probability distribution fluctuation value of the current moment comprises the following steps:

5.1 Determining a probability distribution statistical analysis boundary:

/>

in the method, in the process of the invention,representative of a vibration data sample obtained during a current data acquisition period and after an amplitude normalization operation; theta (theta) _i Representative vibration data sample->Model parameters of (2); />Representing vibration data samples collected in a normal operation stage of the historical wind driven generator and subjected to amplitude standardization operation; theta (theta) ₀ Representative vibration data sample->Is used for the model parameters of the model. l (L) _R 、l _L The upper and lower boundaries of the probability distribution model, respectively. Δμ is a correction term, and the calculation method is as follows:

5.2 Calculating the fluctuation value of the probability distribution at the current moment:

6) And determining a threshold range of the probability distribution fluctuation value, and judging the running state of the wind driven generator at the current moment according to the calculated probability distribution fluctuation value.

The method for determining the threshold range of the probability distribution fluctuation value comprises the following steps of:

6.1 N vibration data samples of the normal operation stage of the wind driven generator are continuously collected, and corresponding probability distribution fluctuation values are calculated respectively.

6.2 Respectively calculating the mean value and standard deviation of the N probability distribution fluctuation values, thereby setting a 3 sigma threshold interval.

6.3 Aiming at vibration data of the running process of the wind driven generator which is continuously collected, calculating probability distribution fluctuation values of the vibration data by using a formula (16), and if the probability distribution fluctuation values do not exceed a 3 sigma threshold interval, indicating that the running state of the wind driven generator is normal; otherwise, the running process is abnormal.

And verifying vibration data of the wind driven generator in the running process of the wind driven generator judged to be in an abnormal state by adopting a spectrum analysis method, and if the fault characteristic frequency is obvious, indicating that the wind driven generator has faults.

Example 23:

referring to fig. 1 to 7, a wind driven generator variable working condition state monitoring method based on amplitude standardization statistical analysis comprises the following steps:

step 1: vibration data in the running process are collected by using a sensor arranged on the wind driven generator.

Step 2: and establishing an amplitude standardization strategy by determining a reference sample and calculating a statistical characteristic ratio coefficient according to vibration data of the running process of the wind driven generator acquired at the current moment.

Step 3: and carrying out amplitude standardization on vibration data of the running process of the wind driven generator, which is acquired at the current moment, according to the established amplitude standardization strategy.

Step 4: and establishing a probability distribution model for describing the vibration data statistical characteristics of the wind driven generator, and determining parameters in the characteristic model by adopting a maximum log likelihood estimation method.

Step 5: determining a probability distribution statistical analysis boundary, comparing and analyzing the probability distribution of vibration data of the wind driven generator operation process obtained at the current moment with the reference probability distribution of the normal operation stage, and calculating the fluctuation value of the probability distribution at the current moment.

Step 6: and determining a threshold range of the probability distribution fluctuation value, and judging the running state of the wind driven generator at the current moment according to the calculated probability distribution fluctuation value.

The step 2 specifically comprises the following steps:

step 201: vibration data x of wind driven generator operation process acquired at current moment (set as ith acquisition period) _i Calculate a reference sample x _b ：

Wherein L represents the sample length of vibration data, x _i-1 Vibration data, x, of the running process of the wind driven generator, which are acquired in the (i-1) th acquisition period _r For vibration data samples of a preselected normal operation phase of the wind turbine, t _i And t _M Representing the time of the current acquisition period and the maintenance time of the wind turbine, respectively.

Step 202: calculating a statistical feature ratio coefficient:

in the method, in the process of the invention,represents a statistical characteristic function omega _i (t) and ω _b (t) rotational speed information representing the current acquisition period and the reference sample acquisition period, respectively.

Step 203: in order to eliminate the influence of abnormal interference and improve the robustness of amplitude normalization, the statistical feature ratio coefficient is further set as:

wherein K represents the total number of samples adjacent to the reference sample and having a statistical characteristic value exceeding the statistical characteristic of vibration data of the ith acquisition period. b is the determined reference sample number. k is the vibration data sample number adjacent to the reference sample.

The step 3 specifically comprises the following steps:

step 301: the method for carrying out amplitude standardization on vibration data of the running process of the wind driven generator acquired at the current moment according to the established amplitude standardization strategy comprises the following steps:

in the method, in the process of the invention,representing wind turbine operational process vibration data after the amplitude normalization operation.

The step 4 specifically comprises the following steps:

step 401: establishing a probability distribution model for describing the vibration data statistical characteristics of the wind driven generator:

wherein, alpha is a characteristic index, beta is a symmetrical parameter, gamma is a scale parameter, delta is a position parameter, and j is an imaginary unit. sgn (t) is a sign function, and the calculation formula is:

in addition, the parameter t represents a regression coefficient, which can be obtained by an interpolation method:

where Λ (·) represents the interpolation function and ζ is the step size.

Step 402: establishing a maximum log-likelihood function:

in the method, in the process of the invention,and a probability density function representing vibration data of the wind driven generator running process after the amplitude normalization operation, wherein Θ= { alpha, beta, gamma, delta } is a model parameter.

Step 403: calculation by numerical integration method

Wherein z is _n The representative sample data centralization operation is as follows:

in addition, the calculation method of v (θ) in the formula (9) is:

wherein, the value range of theta is [0, pi/2 ].

Step 404: estimating the model parameters Θ= { α, β, γ, δ }, by maximizing the log-likelihood function:

the step 5 specifically comprises the following steps:

step 501: determining a probability distribution statistical analysis boundary:

/>

in the method, in the process of the invention,representative of a vibration data sample obtained during a current data acquisition period and after an amplitude normalization operation; theta (theta) _i Representative vibration data sample->Model parameters of (2); />Representing vibration data samples collected in a normal operation stage of the historical wind driven generator and subjected to amplitude standardization operation; theta (theta) ₀ Representative vibration data sample->Is used for the model parameters of the model. l (L) _R 、l _L The upper and lower boundaries of the probability distribution model, respectively. Δμ is a correction term, and the calculation method is as follows:

step 502: calculating the fluctuation value of the probability distribution at the current moment:

The step 6 specifically comprises the following steps:

step 601: n vibration data samples of the normal operation stage of the wind driven generator are continuously collected, and corresponding probability distribution fluctuation values are calculated respectively.

Step 602: the mean and standard deviation of the N probability distribution fluctuation values are calculated, respectively, thereby setting a 3σ threshold interval.

Step 603: calculating probability distribution fluctuation values of vibration data of the wind driven generator in the running process of the subsequent continuous collection by using a formula (16), and if the probability distribution fluctuation values do not exceed a 3 sigma threshold interval, indicating that the running state of the wind driven generator is normal; otherwise, the running process is abnormal.

Step 604: and verifying vibration data of the wind driven generator in the running process of the wind driven generator judged to be in an abnormal state by adopting a spectrum analysis method, and if the fault characteristic frequency is obvious, indicating that the wind driven generator has faults.

Example 24:

referring to fig. 1 to 7, a wind driven generator variable working condition state monitoring method based on amplitude standardization statistical analysis mainly comprises the following technical contents:

the embodiment is developed based on actual measurement data of a 2MW wind driven generator in a certain wind field of the North plain so as to verify the effectiveness of the wind driven generator variable-working-condition state monitoring method based on the amplitude standardized statistical analysis in practical industrial application.

The method comprises the following steps:

step 1: vibration data in the running process are collected by using a sensor arranged on the wind driven generator. In the process of test development, a data acquisition unit (Data acquisition unit, DAU) suitable for a wind driven generator running state monitoring scene is arranged on the driving end and the non-driving end so as to acquire vibration signals of bearings at two ends. In the data acquisition process, the sampling frequency is set to 25.6kHz, the single sampling period is set to 1s, the start-stop time of data acquisition is from 5 months in 2018 to 11 months in 2018, and the bearing data of the driving end of the wind driven generator, which are repeated, incomplete and invalid, are removed for 223 days in total, can be analyzed, as shown in fig. 2, and the corresponding rotating speed information of the main shaft of the wind driven generator in the vibration data acquisition process of the bearing of the driving end of the wind driven generator is shown in fig. 3.

Step 2: and determining a reference sample and calculating a statistical characteristic ratio coefficient by adopting the method according to vibration data of the running process of the wind driven generator acquired at the current moment, and establishing an amplitude standardization strategy on the basis.

Step 3: and (3) carrying out amplitude standardization on vibration data of the wind driven generator in the running process acquired at the current moment according to the amplitude standardization strategy established in the step (2).

Step 4: and establishing a probability distribution model for describing the vibration data statistical characteristics of the wind driven generator, and determining parameters in the characteristic model by adopting a maximum log likelihood estimation method, so that probability distribution corresponding to the vibration data of the bearing at the driving end of the wind driven generator acquired in the current period can be obtained quantitatively.

Step 5: and determining a probability distribution statistical analysis boundary, comparing and analyzing the probability distribution of vibration data of the wind driven generator operation process obtained at the current moment with the reference probability distribution of the normal operation stage, and calculating the fluctuation value of the probability distribution at the current moment, wherein the result is shown in figure 4.

Step 6: and determining a threshold range of the probability distribution fluctuation value, and judging the running state of the wind driven generator at the current moment according to the calculated probability distribution fluctuation value. The probability distribution fluctuation values of vibration data of the driving end bearing of the wind driven generator in the first 50 normal operation stages are adopted, the mean value and the standard deviation of the vibration data are calculated, so that a 3 sigma threshold interval is set, the threshold interval obtained in the test is [ -0.010,0.014], as shown in fig. 4, and therefore, the driving end bearing of the wind driven generator can be judged to start to be obviously abnormal in the 126 th data acquisition period, and the driving end bearing can be indicated to possibly have early faults at the moment. To verify the accuracy of this determination, the 126 th data acquisition sample was analyzed using a spectral analysis method, the results of which are shown in fig. 5. It can be seen that in the data acquisition period, the inner ring fault characteristic frequency fi, the outer ring fault characteristic frequency fo and the rolling body fault characteristic frequency fb of the driving end bearing are highlighted, which indicates that the driving end bearing is indeed in fault at the moment, so that the effectiveness of the wind driven generator variable working condition state monitoring method based on the amplitude standardized statistical analysis is also verified.

In addition, in order to compare the performance of the method provided by the present invention, the state monitoring methods that are currently more commonly used, such as Root Mean Square (RMS), activity, mobility, average info gram, zhan Senrui Li Sandu (MFEn-IJRD) based on multi-scale fuzzy entropy, spectrum Maximum (MAXS), spectrum Variance (VARS), generalized Expansion Hjorth Parameter (GEHP), and the like, are compared and analyzed, and the result is shown in fig. 6.

As can be seen from FIG. 6, the degradation curve of the bearing at the driving end of the wind driven generator obtained by the method is smoother, the data fluctuation caused by the variable working condition is effectively restrained, and the problems of false alarm or missing alarm of the state and the like possibly caused by the data fluctuation are reduced. The other comparison methods have serious fluctuation, and the depicting capability of the bearing degradation process at the driving end of the wind driven generator is insufficient. By setting the corresponding threshold interval, the abnormal initial points detected by different methods are shown in table 1, and the method has better early fault sensitivity.

TABLE 1 comparison of abnormal initial points detected by different methods

In addition to the analysis of the abnormal initial points, the present embodiment also analyzed the trend and the identifiability of the bearing degradation curve of the driving end of the wind turbine obtained by the different methods, and the results are shown in table 2.

In order to more objectively compare the early abnormality detection performance of the bearing at the driving end of the wind turbine with the above-mentioned different methods, the present embodiment uses a test subject working characteristic (Receiver operating characteristic, ROC) curve for analysis, and the results are shown in fig. 7.

TABLE 2 trends and recognizabilities of degradation curves obtained by different methods

The method can still obtain higher abnormality detection accuracy under the condition of lower false alarm rate, and the effectiveness of the method in the condition of monitoring the state of the wind driven generator is verified.

Claims (10)

1. The wind driven generator variable working condition state monitoring method based on the amplitude standardization statistical analysis is characterized by comprising the following steps of:

1) Collecting vibration data in the running process of the wind driven generator in real time;

2) And calculating a statistical feature ratio coefficient according to vibration data acquired in real time.

3) And carrying out amplitude standardization processing on vibration data acquired in real time according to the statistical characteristic ratio coefficient to obtain a vibration data sample with standardized amplitude.

4) Establishing a probability distribution model for describing the vibration data statistical characteristics of the wind driven generator, and determining parameters in the probability distribution model;

5) Inputting the vibration data sample with standardized amplitude into a probability distribution model to obtain the probability distribution of vibration data in the running process of the wind driven generator;

6) Comparing and analyzing the vibration data probability distribution of the wind driven generator operation process obtained at the current moment with the reference probability distribution of the normal operation stage, and calculating the probability distribution fluctuation value of the vibration data sample with standardized amplitude at the current moment;

7) Judging whether the fluctuation value exceeds a preset threshold range, if so, judging that the current wind driven generator operates abnormally, and if not, the wind driven generator operates normally.

2. The method for monitoring the variable working condition state of the wind driven generator based on the amplitude standardization statistical analysis of claim 1, wherein vibration data in the running process of the wind driven generator is collected through a sensor arranged on the wind driven generator.

3. The method for monitoring the variable working condition state of the wind driven generator based on the amplitude standardization statistical analysis according to claim 1, wherein the step of calculating the statistical feature ratio coefficient according to the vibration data collected in real time comprises the following steps:

2.1 Calculating a reference sample x _b (n) as follows:

wherein i is the acquisition period sequence number, x _i-1 (n) vibration data of the running process of the wind driven generator, which are acquired in the (i-1) th acquisition period; x is x _r (n) is a vibration data sample of the normal operation stage of the historical wind driven generator; n is the sample length number of the vibration data; l represents the total length of the sample of vibration data; t is t _i A time representing a current acquisition period; t is t _M Representing maintenance time of the wind driven generator;

2.2 Calculating a statistical feature ratio coefficient as follows:

in the method, in the process of the invention,is a statistical feature ratio coefficient; x is x _i (n) vibration data of the running process of the wind driven generator, which are acquired in the ith acquisition period; omega _i (t) rotational speed information representing an i-th acquisition period; omega _b (t) rotational speed information representing a reference sample acquisition period; />Representing a statistical feature function; k represents vibration close to the reference sample and the statistical characteristic values exceed the ith acquisition periodCounting the total number of samples of the features; b is a reference sample number; k is the vibration data sample number adjacent to the reference sample.

4. The method for monitoring the variable working condition state of the wind driven generator based on the amplitude standardization statistical analysis of claim 1, wherein the amplitude standardization vibration data sample is as follows:

wherein n is the sample length number of the vibration data; phi [ x ] _i (n),x _b (n),t _M ,K]Is a statistical feature ratio coefficient; x is x _i (n) vibration data of the running process of the wind driven generator, which are acquired in the ith acquisition period;representative of a vibration data sample of normalized amplitude.

5. The method for monitoring the variable working condition state of the wind driven generator based on the amplitude standardization statistical analysis according to claim 1, wherein the probability distribution model for describing the vibration data statistical characteristics of the wind driven generator is as follows:

Wherein phi (t) is the probability distribution of vibration data in the running process of the wind driven generator; alpha is a characteristic index, beta is a symmetrical parameter, gamma is a scale parameter, delta is a position parameter, and j is an imaginary unit; t is a regression coefficient;

wherein the sign function sgn (t) is as follows:

6. the method for monitoring the variable working condition state of the wind driven generator based on the amplitude standardization statistical analysis according to claim 5, wherein the regression coefficient t is as follows:

wherein, Λ (·) represents an interpolation function, and ζ is a step size; n is the sample length number of the vibration data;representative of a vibration data sample of normalized amplitude.

7. The method for monitoring the variable working condition state of the wind driven generator based on the amplitude standardization statistical analysis according to claim 1, wherein the step of determining parameters in the probability distribution model comprises the following steps:

4.1 Establishment of maximum log likelihood functionThe following is shown:

wherein n is the sample length number of the vibration data; l represents the total length of the sample of vibration data;probability density function representing amplitude normalized vibration data samples, +.>Vibration data sample representing normalized amplitudeThe cost is high; Θ is a model parameter, Θ= { α, β, γ, δ }, α is a feature index, β is a symmetry parameter, γ is a scale parameter, and δ is a position parameter;

4.2 Calculating a probability density function for sample data centralization by a numerical integration methodThe following is shown:

wherein the sample data z is centralized _n The following is shown:

the parameter v (θ) is as follows:

wherein, theta is a variable, and the value range is [0, pi/2 ];

4.3 Estimating the model parameters Θ by means of a maximum log-likelihood function, as follows:

8. the method for monitoring the variable working condition state of the wind driven generator based on the amplitude standardization statistical analysis according to claim 1, wherein the step of comparing the probability distribution of vibration data of the wind driven generator operation process obtained at the current moment with the reference probability distribution of the normal operation stage and calculating the probability distribution fluctuation value of the vibration data sample with the standardized amplitude at the current moment comprises the following steps:

6.1 Determining the boundaries of the probability distribution model as follows:

in the method, in the process of the invention,a vibration data sample representative of an amplitude normalization of the ith acquisition period; />A vibration data sample representing the amplitude standardization of the normal operation stage of the historical wind driven generator; theta (theta) _i Representing vibration data +.>Model parameters of (2); theta (theta) ₀ Representative vibration data sample->Model parameters of (2); l (L) _R 、l _L The upper boundary and the lower boundary of the probability distribution model are respectively;

Wherein the correction term Δμ is as follows:

6.2 Calculating the probability distribution fluctuation value of the vibration data sample with the normalized amplitude at the current moment, as follows:

in the method, in the process of the invention,the probability distribution fluctuation value of the vibration data sample normalized for the amplitude of the current moment.

9. The method for monitoring the variable working condition state of the wind driven generator based on the amplitude standardization statistical analysis according to claim 1, wherein the step of setting the threshold range comprises the following steps:

7.1 Continuously collecting N vibration data samples of the wind driven generator in a normal operation stage, and respectively calculating corresponding probability distribution fluctuation values;

7.2 Calculating the mean value and standard deviation of probability distribution fluctuation values of the N vibration data samples;

7.3 A 3 sigma threshold interval is set according to the mean value and standard deviation of probability distribution fluctuation values of the N vibration data samples.

10. The wind driven generator variable working condition state monitoring method based on amplitude standardization statistical analysis according to claim 1, wherein if the current wind driven generator is judged to be abnormal in operation, a spectrum analysis method is adopted for verification;

if the fault characteristic frequency is obvious, the wind driven generator fails.