A method for evaluating equipment anomalies under extreme weather conditions

Abstract

The present invention relates to the field of wind rotor abnormality detection, and specifically to a method for evaluating equipment abnormality under extreme weather conditions. First, signals of various frequencies in a vibration signal are separated by filtering to obtain a fundamental frequency signal and a 3-fold frequency signal. By comparing the amplitudes of the two, a first fault signal is extracted from the fundamental frequency signal and the 3-fold frequency signal. Then, the complexity of the first signal at the moment corresponding to the first fault signal is further calculated, thereby realizing the extraction of a second fault signal. Finally, a failure rate function model is established to calculate the probability of the second fault signal occurring within a time range, and the probability is compared with a failure rate threshold to determine whether a wind rotor of a wind turbine has failed. This method can automatically and quickly identify whether the wind rotors of all wind turbines in a power station are loose, without the need to inspect each wind rotor in turn. This solves the problem of slow progress when the status of wind rotors is currently detected in a semi-manual manner.

Description

Technical Field

[0001] The present invention relates to the field of wind wheel anomaly detection, and in particular to a method for evaluating equipment anomaly under extreme weather conditions. Background Art

[0002] In extremely windy weather, the strong wind load will exert great pressure on the surface of the wind rotor, causing damage to the main gear, gearbox and bolts inside the wind rotor, thereby causing the wind rotor to loosen and vibrate abnormally. The abnormal vibration will further increase the falling speed of bolts and other parts, affecting the safe operation of the wind rotor.

[0003] In the existing technology, when inspecting the status of wind rotors after extremely windy weather, the inspection is generally carried out in a semi-manual manner, in which the bolts and other parts on the surface of the wind rotor are inspected to check the tightness of the wind rotor. External force is applied manually or various instruments are used to check whether the parts are damaged or loose. However, this method requires each wind rotor to be inspected in turn during the inspection, which is relatively slow and difficult to quickly complete the inspection of all wind rotors in the smart power station, resulting in a long inspection cycle. This will also affect the time it takes for the wind power station to be put into use after strong winds and extreme weather, affecting the efficiency of the power station. Summary of the Invention

[0004] In response to the deficiencies of the prior art, the present invention provides a method for evaluating equipment anomalies under extreme weather conditions, which solves the problem of slow progress when the status of wind wheels is currently detected sequentially in a semi-manual manner.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] A method for evaluating equipment abnormality under extreme weather conditions, the method comprising the following steps:

[0007] S1. Obtaining a vibration signal of a wind turbine rotor including axial and radial vibrations;

[0008] S2. Filtering the vibration signal to separate the fundamental signal and the harmonic signal in the vibration signal, and obtaining the amplitudes of signals of different frequencies to obtain a first signal;

[0009] S3. Compare the amplitudes of the baseband signal and the tripled frequency signal in the first signal to obtain a first fault signal;

[0010] S4. Calculate the complexity of the first signal at a time corresponding to the first fault signal, and select the first fault signal corresponding to a time when the complexity is greater than a complexity threshold to generate a second fault signal;

[0011] S5. Obtain a spanning time range of the second fault signal, and set a failure rate function model to calculate the failure rate of the second fault signal within the spanning time range;

[0012] S6. Determine whether a wind turbine rotor fails based on whether the total failure rate of the second failure signals during the extreme weather period exceeds a failure rate threshold;

[0013] If so, output a prompt message indicating that the wind turbine rotor is abnormal;

[0014] If not, a prompt message indicating no abnormality is output.

[0015] Preferably, in step S2, the following steps are specifically included:

[0016] S21. Establish a Kalman filter system model;

[0017] S22. Setting a likelihood function for optimizing the model parameters of the Kalman filter system model, and obtaining the optimal parameters of the Kalman filter system model according to maximum likelihood estimation; the optimal parameters satisfy the following relationship:

[0018]

[0019] In the above formula, μ * represents the optimal parameters of the Kalman filter system model obtained when the likelihood function takes the minimum value, μ represents the model parameter set of the Kalman filter system model, represents the covariance matrix of the estimation error, represents the noise matrix, c1 represents the sampling length of the Kalman filter system model;

[0020] S23, substituting the obtained optimal parameters into the Kalman filter system model, and filtering the vibration signal to obtain an optimal state matrix of the filtered vibration signal;

[0021] S24. Obtain the amplitudes of different frequency signals at different times according to the optimal state matrix of the vibration signal to obtain a first signal; the expression of the first signal is:

[0022]

[0023] In the above formula, A i，k and The frequencies of the vibration signals at time k are The amplitude and initial phase of the signal component, x 2i，k Represents the 2i-th number in the optimal state matrix at time k.

[0024] Preferably, in step S3, the following steps are specifically included:

[0025] S31, obtaining a baseband signal and a triple frequency signal from a first signal;

[0026] S32, setting a fundamental frequency amplitude threshold, and determining whether the fundamental frequency amplitude of the fundamental frequency signal is greater than a first amplitude threshold;

[0027] If yes, extract the fundamental frequency signal whose fundamental frequency amplitude is greater than the first amplitude threshold and the triple frequency signal at the corresponding moment to obtain the second signal;

[0028] If not, then end;

[0029] S33, setting the basic multiple, and calculating the second amplitude of the tangential 3-fold frequency amplitude; the calculation formula of the second amplitude is:

[0030]

[0031] In the above formula, The second amplitude of the tangential triple frequency amplitude, A 3 represents the tangential triple frequency amplitude, and τ represents the basic multiple;

[0032] S34, determining whether the tangential fundamental frequency amplitude in the second signal is greater than the second amplitude of the tangential triple frequency amplitude;

[0033] If yes, extract the second signal at the corresponding moment to obtain the third signal;

[0034] If not, then end;

[0035] S35. Calculate the duration of each baseband signal and triple frequency signal in the third signal. The calculation formula for the duration is:

[0036] Δt(i)=t 1 (i)-t 0 (i)

[0037] In the above formula, Δt(i) represents the duration of the i-th third signal, t 0 (i) and t 1 (i) represents the initial time and the final time of the i-th third signal respectively;

[0038] S36 , setting a duration threshold, and selecting a third signal whose duration is greater than the duration threshold to generate a first fault signal.

[0039] Preferably, in step S4, the following steps are specifically included:

[0040] S41, marking the time corresponding to the first fault signal as a reference time;

[0041] S42. Set a number of amplitude intervals, calculate the probability of the first signal corresponding to the amplitude interval at any reference time, that is, the first amplitude probability, and normalize the first amplitude probability to obtain a second amplitude probability;

[0042] S43. Calculate the complexity of the first signal at any moment according to the amplitude probability;

[0043] S44 , identifying a moment when the complexity is greater than a complexity threshold, and selecting a first fault signal corresponding to the moment to obtain a second fault signal.

[0044] Preferably, in step S42, the following steps are specifically included:

[0045] S421, setting a plurality of amplitude intervals, and making the amplitude value of each moment in the first signal correspond to each amplitude interval;

[0046] S422. Calculate the probability of the first signal corresponding to the amplitude interval at any time, that is, the first amplitude probability. The calculation formula for the first amplitude probability is:

[0047]

[0048] In the above formula, It represents the probability of the first signal appearing in the corresponding amplitude interval at the rth reference moment, that is, the first amplitude probability, N r N represents the number of amplitude intervals corresponding to the first fault signal at the rth reference moment in the first signal, R represents the total number of amplitude intervals in the first signal;

[0049] S423. Normalize the first amplitude probability to obtain a second amplitude probability of the first fault signal appearing in the amplitude interval corresponding to the first fault signal at any time. The calculation formula for the second amplitude probability is:

[0050]

[0051] In the above formula, represents the probability of the second amplitude occurring in the amplitude interval corresponding to the first signal at the rth moment, It represents the first amplitude probability of the first signal corresponding to the amplitude interval at the rth moment, and the total number of amplitude intervals on the first signal is q.

[0052] Preferably, in step S5, the following steps are specifically included:

[0053] S51, obtaining a span duration range of the second fault signal from the first moment to the last moment;

[0054] S52, marking moments that do not correspond to the second fault signal within the spanning time range as blank moments, and marking several adjacent blank moments as blank periods;

[0055] S53. Setting a failure rate function model in which the failure rate changes with time. The failure rate function model satisfies the following relationship:

[0056]

[0057] In the above formula, f(t i ) indicates that at t i The output of the failure rate function model at time K i Indicates that at t i The slope of the failure rate function model at time K 0 represents the upper limit of the slope;

[0058] S54. According to the failure rate function model, the function value at the time corresponding to the second fault signal is taken as a positive number, and the function value at the time corresponding to the blank period is taken as a negative number, and the total failure rate of the second fault signal within the span time range is calculated.

[0059] Preferably, in step S52, the following steps are specifically included:

[0060] S521, setting a time window;

[0061] S522: Obtain a moment in the spanning time range that does not correspond to the second fault signal, and mark it as a blank moment;

[0062] S523, traversing each blank moment in sequence through the time window, and determining whether the moments corresponding to both ends of the time window are the moments corresponding to the second fault signal;

[0063] If yes, then mark the time corresponding to the time window as the time corresponding to the second fault signal;

[0064] If not, proceed to the next step;

[0065] S524: Mark several adjacent blank moments as a blank period.

[0066] Preferably, in step S54, the following steps are specifically included:

[0067] S541. Set several time intervals and set weight coefficients corresponding to the time intervals;

[0068] S542. Calculate the filling amount of each blank moment and the second fault signal within the span duration range according to the weight coefficient. The calculation formula of the filling amount is:

[0069]

[0070] In the above formula, W i Indicates the filling amount of the i-th blank period or the second fault signal within the time range, ti1 and t im represents the first and last moments of the i-th blank period within the duration range, σ i Represents the time interval [t i1 , t im ]The corresponding weight coefficient, f(x) represents the failure rate function model;

[0071] S543: Calculate the total failure rate of the second fault signal within the span time range according to the filling amount.

[0072] Preferably, in step S43, the complexity calculation formula is:

[0073]

[0074] In the above formula, FZ r represents the complexity of the first signal at the rth moment, is the probability of the second amplitude occurring in the amplitude interval corresponding to the rth moment at the Fth frequency. There are a total of F1 frequencies at the rth moment in the first signal.

[0075] Preferably, in step S543, the total failure rate is calculated as follows:

[0076]

[0077] In the above formula, P represents the total failure rate of the second fault signal within the span time range, W i and W j They represent the filling amount of the i-th second fault signal and the filling amount of the j-th blank period within the time range, T 1 represents the set of second fault signals, T 2 Represents a set of blank periods, t1 to t m That is, the first moment and the last moment of the span duration, σ0 represents the weight coefficient corresponding to the span duration, and f(x) represents the failure rate function model.

[0078] Compared with the existing technology, the present invention provides a method for evaluating equipment anomalies in extreme weather conditions, which has the following beneficial effects:

[0079] 1. First, the vibration signal of the wind rotor is obtained by arranging a vibration sensor, and then the signals of each frequency in the vibration signal are filtered and separated to obtain the base frequency signal and the 3rd frequency signal. By comparing the amplitudes of the two, the first fault signal is extracted from the base frequency signal and the 3rd frequency signal. Then, the complexity of the moment corresponding to the first fault signal in the first signal is further calculated to realize the extraction of the second fault signal. Finally, a failure rate function model is established to calculate the probability of the second fault signal occurring within the time range, and compared with the failure rate threshold to determine whether the wind rotor of the wind turbine generator has a fault. This method can automatically and quickly identify whether the wind rotors of all wind turbine generators in the power station are loose, without the need to check each wind rotor in turn, which is more convenient and quick.

[0080] 2. The present invention obtains the second signal by extracting the fundamental frequency signal and the triple frequency signal from the first signal, and extracting the fundamental frequency signal and the triple frequency signal at the moment when the fundamental frequency amplitude is greater than the first amplitude threshold. Then, the third signal is obtained by comparing the tangential fundamental frequency amplitude and the tangential triple frequency amplitude, and then the noise is filtered through the duration threshold to select the first fault signal.

[0081] 3. The present invention sets amplitude intervals and corresponds different amplitudes on the first signal to each amplitude interval, thereby calculating the first amplitude probability and the second amplitude probability of the amplitude interval and finally calculating the complexity, thereby achieving further screening of the first fault signal to obtain a more accurate second fault signal, thereby improving the accuracy of the subsequent total failure rate.

[0082] 4. By setting different time intervals and corresponding weight coefficients, the present invention can assign a larger weight coefficient to the continuous area and blank period corresponding to the second fault signal that spans a longer time period when calculating the filling amount, thereby improving its credibility, increasing the impact on the total failure rate, and improving the accuracy of the total failure rate.

[0083] 5. The present invention sets a specific failure rate function model. Considering that the end of the spanning time range is the end of the extreme windy weather, after the wind wheel has experienced a certain period of extreme windy weather, the state near the end of the spanning time range can better reflect its true state after the extreme windy weather, thereby increasing its influence on the total failure rate. At the same time, in order to avoid excessive influence, an upper limit of the slope is also set to limit it, thereby achieving the purpose of more accurately calculating the total failure rate of the wind wheel. BRIEF DESCRIPTION OF THE DRAWINGS

[0084] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0085] Figure 1 This is a flow chart of the method for evaluating equipment abnormality under extreme weather conditions of the present invention;

[0086] Figure 2 Schematic diagram of blank period and time window of the present invention;

[0087] Figure 3 This is a schematic diagram of the present invention after the blank period is replaced by the moment corresponding to the second fault signal. DETAILED DESCRIPTION

[0088] To make the above-mentioned objectives, features, and advantages of the present invention more clearly understood, the present invention is further described below in detail with reference to the accompanying drawings and specific embodiments. This will enable a full understanding of how this application uses technical means to solve technical problems and achieve technical effects, and to implement the invention accordingly.

[0089] Those skilled in the art will appreciate that all or part of the steps in the following embodiments can be accomplished by instructing related hardware through a program. Therefore, the present application may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present application may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0090] In order to solve the problem of slow progress when the status of wind rotors is detected in sequence by semi-manual means, the present invention provides a method for evaluating equipment abnormality in extreme weather conditions. The method can evaluate whether the wind rotors are loose during the entire process of strong winds and extreme weather conditions, and directly output the results of whether each wind rotor is loose after the strong winds, so that the operation and maintenance personnel can directly perform maintenance on the corresponding wind rotors more quickly without having to check whether each wind rotor is loose. Figure 1 As shown, the method includes the following steps:

[0091] S1. Obtain vibration signals of a wind turbine rotor including axial and radial vibrations. The vibration signals can be obtained by installing a vibration sensor on the rotor, generally on a bearing of the rotor.

[0092] S2. Filter the vibration signal to separate the fundamental signal and harmonic signal in the vibration signal, and obtain the amplitude of signals of different frequencies to obtain a first signal. Since the wind rotor as a whole consists of multiple parts, each part may vibrate strongly in windy weather. Therefore, the detected vibration signal may contain various frequencies. Therefore, it is necessary to separate and extract the signals of each frequency to obtain the signal of the required frequency for subsequent calculation. Therefore, in step S2, the following steps are specifically included:

[0093] S21. Establish a Kalman filter system model. Here, it is assumed that the discrete form of the wind wheel vibration signal is:

[0094]

[0095] Based on this, the state matrix of the designed Kalman filter system model is:

[0096]

[0097] The observation matrix is:

[0098] H k =[coswkT S ,-sinwkT S ,...,cosnwkT S , -sinnwkT S ] I×2n

[0099] In the above formula, y k Represents the vibration signal, T S Indicates the duration of the sampling time, n indicates the highest harmonic, A i，k and The frequencies at time k are The amplitude and initial phase of the signal component, e(k is the measurement noise, is the state matrix;

[0100] S22. Setting a likelihood function for optimizing the model parameters of the Kalman filter system model, and obtaining the optimal parameters of the Kalman filter system model according to maximum likelihood estimation; the optimal parameters satisfy the following relationship:

[0101]

[0102] In the above formula, μ * represents the optimal parameters of the Kalman filter system model obtained when the likelihood function takes the minimum value, μ represents the model parameter set of the Kalman filter system model, represents the covariance matrix of the estimation error, represents the noise matrix, c1 represents the sampling length of the Kalman filter system model;

[0103] S23, substituting the obtained optimal parameters into the Kalman filter system model, and filtering the vibration signal to obtain an optimal state matrix of the filtered vibration signal;

[0104] S24. Obtain the amplitudes of different frequency signals at different times according to the optimal state matrix of the vibration signal to obtain a first signal; the expression of the first signal is:

[0105]

[0106] In the above formula, A i，k and The frequencies of the vibration signals at time k are The amplitude and initial phase of the signal component, x 2i，k Represents the 2i-th number in the optimal state matrix at time k.

[0107] S3. Compare the amplitudes of the fundamental frequency signal and the triple frequency signal in the first signal to obtain a first fault signal. After a large number of experiments, it is found that when the wind rotor is loose and vibrates, there is a certain relationship between the amplitudes of the fundamental frequency signal and the triple frequency signal, thereby achieving a preliminary judgment on whether the wind rotor is vibrating. Therefore, in step S3, the following steps are specifically included:

[0108] S31, obtaining a baseband signal and a triple frequency signal from a first signal;

[0109] S32, setting a fundamental frequency amplitude threshold, and determining whether the fundamental frequency amplitude of the fundamental frequency signal is greater than a first amplitude threshold;

[0110] If yes, extract the fundamental frequency signal whose fundamental frequency amplitude is greater than the first amplitude threshold and the triple frequency signal at the corresponding moment to obtain the second signal;

[0111] If not, then end;

[0112] S33, setting the basic multiple, and calculating the second amplitude of the tangential 3-fold frequency amplitude; the calculation formula of the second amplitude is:

[0113]

[0114] In the above formula, The second amplitude of the tangential triple frequency amplitude, A 3 It represents the 3rd harmonic amplitude of the tangential direction, and τ represents the fundamental multiple, which is generally 1.5 to 2.2, so as to filter out some noise. Under normal conditions, the amplitude of the fundamental frequency and the 3rd harmonic amplitude of the tangential direction are almost equal.

[0115] S34, determining whether the tangential fundamental frequency amplitude in the second signal is greater than the second amplitude of the tangential triple frequency amplitude;

[0116] If yes, extract the second signal at the corresponding moment to obtain the third signal;

[0117] If not, then end;

[0118] S35. Calculate the duration of each baseband signal and triple frequency signal in the third signal. The calculation formula for the duration is:

[0119] Δt(i)=t 1 (i)-t 0 (i)

[0120] In the above formula, Δt(i) represents the duration of the i-th third signal, t 0 (i) and t 1 (i) represents the initial time and the final time of the i-th third signal respectively;

[0121] S36. Setting a duration threshold and selecting a third signal whose duration is greater than the duration threshold to generate a first fault signal. The duration is set to eliminate noise data that occurs in a very short time.

[0122] S4. Calculate the complexity of the first signal at the time corresponding to the first fault signal, and select the first fault signal corresponding to the time when the complexity is greater than the complexity threshold to generate a second fault signal. In windy weather, the vibration frequency and amplitude of the loose wind wheel are more complex. By calculating the complexity, the first fault signal is secondary screened to obtain the second fault signal. In step S4, the following steps are specifically included:

[0123] S41. Mark the time corresponding to the first fault signal as the reference time. The reference time is set to reduce the amount of computational complexity. Only the complexity corresponding to the reference time needs to be calculated.

[0124] S42. Set several amplitude intervals, and calculate the probability of the first signal corresponding to the amplitude interval at any reference time, that is, the first amplitude probability, and normalize the first amplitude probability to obtain the second amplitude probability. The amplitudes of different frequencies vary. To facilitate the subsequent calculation of the complexity at any time, it is necessary to divide the different amplitudes into intervals, and replace the amplitude of the first signal at any time with the amplitude interval to facilitate the calculation of the first amplitude probability and the second amplitude probability. Therefore, in step S42, the following steps are specifically included:

[0125] S421, setting a plurality of amplitude intervals, and making the amplitude value of each moment in the first signal correspond to each amplitude interval;

[0126] S422. Calculate the probability of the first signal corresponding to the amplitude interval at any time, that is, the first amplitude probability. The calculation formula for the first amplitude probability is:

[0127]

[0128] In the above formula, It represents the probability of the first signal appearing in the corresponding amplitude interval at the rth reference moment, that is, the first amplitude probability, N r N represents the number of amplitude intervals corresponding to the first fault signal at the rth reference moment in the first signal, R represents the total number of amplitude intervals in the first signal;

[0129] S423. Normalize the first amplitude probability to obtain a second amplitude probability of the first fault signal appearing in the amplitude interval corresponding to the first fault signal at any time. The calculation formula for the second amplitude probability is:

[0130]

[0131] In the above formula, represents the probability of the second amplitude occurring in the amplitude interval corresponding to the first signal at the rth moment, It represents the first amplitude probability of the first signal corresponding to the amplitude interval at the rth moment, and the total number of amplitude intervals on the first signal is q.

[0132] S43. Calculate the complexity of the first signal at any time according to the amplitude probability; the calculation formula of the complexity is:

[0133]

[0134] In the above formula, FZ r represents the complexity of the first signal at the rth moment, is the probability of the second amplitude appearing in the amplitude interval corresponding to the rth moment at the Fth frequency. There are F1 frequencies in the first signal at the rth moment.

[0135] S44 , identifying a moment when the complexity is greater than a complexity threshold, and selecting a first fault signal corresponding to the moment to obtain a second fault signal.

[0136] S5. Obtain the spanning time range of the second fault signal, and set a failure rate function model to calculate the failure rate of the second fault signal within the spanning time range. The spanning time range is the area between the start time of the first second fault signal and the time of the last second fault signal. Within the spanning time range, there are times corresponding to the second fault signal and times not corresponding to the second fault signal, i.e., blank times. Subsequently, it is necessary to further analyze the total failure rate of the wind rotor based on the times corresponding to the second fault signal and the blank times. In step S5, the following steps are specifically included:

[0137] S51, obtaining a span duration range of the second fault signal from the first moment to the last moment;

[0138] S52, mark the moments that do not correspond to the second fault signal within the span as blank moments, and mark several adjacent blank moments as blank periods; in order to further illustrate the method of obtaining blank moments, and to avoid the moments corresponding to the second fault signal being divided by individual blank periods with extremely short time spans, thereby affecting the subsequent calculation of the total failure rate, as shown in FIG. Figure 2 and Figure 3 As shown, in step S52, the following steps are specifically included:

[0139] S521. Set a time window. The time window is set to prevent occasional calculation errors from causing individual moments to be misjudged as blank moments. It is usually set to several seconds. In the experiment, it is set to 5 seconds.

[0140] S522: Obtain a moment in the spanning time range that does not correspond to the second fault signal, and mark it as a blank moment;

[0141] S523, traversing each blank moment in sequence through the time window, and determining whether the moments corresponding to both ends of the time window are the moments corresponding to the second fault signal;

[0142] If yes, then mark the time corresponding to the time window as the time corresponding to the second fault signal;

[0143] If not, proceed to the next step;

[0144] S524: Mark several adjacent blank moments as a blank period.

[0145] S53. Setting a failure rate function model in which the failure rate changes with time. The failure rate function model satisfies the following relationship:

[0146]

[0147] In the above formula, f(t i ) indicates that at t iThe output of the failure rate function model at time K i Indicates that at t i The slope of the failure rate function model at time K 0 Indicates the upper limit of the slope. Given that the end of the span duration range is the end of the extreme wind weather, the state of the wind rotor near the end of the span duration range after a certain period of extreme wind weather can better reflect its actual state after the extreme wind weather. Therefore, its impact on the total failure rate is increased. At the same time, to avoid excessive impact, an upper limit of the slope is also set to limit it. In practice, we use a straight line with a slope of 5 for calculation.

[0148] S54: Based on the failure rate function model, the function value at the time corresponding to the second fault signal is taken as a positive number, and the function value at the time corresponding to the blank period is taken as a negative number, and the total failure rate of the second fault signal within the span time range is calculated. The blank period indicates that the wind rotor is in a normal state. Therefore, based on the failure rate function model, the ratio of the normal state to the abnormal state is calculated to obtain the total failure rate. In step S54, the following steps are specifically included:

[0149] S541. Set several time intervals and set weight coefficients corresponding to the time intervals. The time intervals correspond to different durations. For example, if the duration of the time interval corresponding to a blank period is longer, it means that the normal probability of the wind turbine is higher. Therefore, different weight coefficients are set to increase the value of the blank period with a longer duration, that is, the filling amount;

[0150] S542. Calculate the filling amount of each blank moment and the second fault signal within the span duration range according to the weight coefficient. The calculation formula of the filling amount is:

[0151]

[0152] In the above formula, W i Indicates the filling amount of the i-th blank period or the second fault signal within the time range, t i1 and t im represents the first and last moments of the i-th blank period within the duration range, σ i Represents the time interval [t i1 , t im ]The corresponding weight coefficient, f(x) represents the failure rate function model;

[0153] S543. Calculate the total failure rate of the second fault signal within the span time range based on the filling amount. The calculation formula for the total failure rate is:

[0154]

[0155] In the above formula, P represents the total failure rate of the second fault signal within the span time range, W i and W j They represent the filling amount of the i-th second fault signal and the filling amount of the j-th blank period within the time range, T 1 represents the set of second fault signals, T 2 Represents a set of blank periods, t1 to t m That is, the first moment and the last moment of the span duration, σ0 represents the weight coefficient corresponding to the span duration, and f(x) represents the failure rate function model.

[0156] S6. Determine whether a wind turbine rotor fails based on whether the total failure rate of the second fault signal during the extreme weather period exceeds a failure rate threshold. In actual operation, the failure rate threshold is set to 0, which is specifically set according to actual needs and is related to the material, process, and size parameters of the wind turbine rotor.

[0157] If so, a prompt message indicating that the wind turbine rotor is abnormal is output; different wind turbine rotors correspond to different numbers, so the actual prompt message will also specify different wind turbine rotor numbers, making it easier for operation and maintenance personnel to directly find the corresponding wind turbine rotor for operation and maintenance;

[0158] If not, a prompt message indicating no abnormality is output.

[0159] When the present invention is implemented, the vibration signal of the wind wheel is first obtained by arranging a vibration sensor, and then the signals of each frequency in the vibration signal are separated by filtering. Then, the base frequency signal and the 3rd frequency signal obtained by separation are compared, and the first fault signal is extracted from the base frequency signal and the 3rd frequency signal. Then, in order to further improve the accuracy of the first fault signal, the complexity of the moment corresponding to the first fault signal in the first signal is further calculated by calculating the complexity, so as to realize the extraction of the second fault signal. Finally, a failure rate function model is established to calculate the probability of the second fault signal failing within a time range, and compared with the failure rate threshold, so as to determine whether the wind wheel of the wind turbine generator has failed. This method can automatically and quickly identify whether the wind wheels of all wind turbines in the power station are loose, and there is no need to check each wind wheel in turn, which is more convenient and quick.

[0160] The above embodiments provide a detailed introduction to the present invention. Specific examples are used herein to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core ideas. At the same time, for those skilled in the art, according to the ideas of the present invention, there may be changes in the specific implementation methods and application scopes. In summary, the contents of this specification should not be understood as limiting the present invention.

Claims

1. A method for evaluating equipment abnormality under extreme weather conditions, characterized in that: The method comprises the following steps: S1. Obtaining a vibration signal of a wind turbine rotor including axial and radial vibrations; S2. Filtering the vibration signal to separate the fundamental signal and the harmonic signal in the vibration signal, and obtaining the amplitudes of signals of different frequencies to obtain a first signal; S3, comparing the amplitudes of the baseband signal and the tripled frequency signal in the first signal to obtain a first fault signal; in step S3, the method specifically includes the following steps: S31, obtaining a baseband signal and a triple frequency signal from a first signal; S32, setting a fundamental frequency amplitude threshold, and determining whether the fundamental frequency amplitude of the fundamental frequency signal is greater than a first amplitude threshold; If yes, extract the fundamental frequency signal whose fundamental frequency amplitude is greater than the first amplitude threshold and the triple frequency signal at the corresponding moment to obtain the second signal; If not, then end; S33, setting the basic multiple, and calculating the second amplitude of the tangential triple frequency amplitude; The calculation formula for the second amplitude is: In the above formula, The second amplitude of the tangential triple frequency amplitude, A 3 represents the tangential triple frequency amplitude, and τ represents the basic multiple; S34, determining whether the tangential fundamental frequency amplitude in the second signal is greater than the second amplitude of the tangential triple frequency amplitude; If yes, extract the second signal at the corresponding moment to obtain the third signal; If not, then end; S35, calculating the duration of each baseband signal and the tripled frequency signal in the third signal; The calculation formula for duration is: Δt(i)=t 1 (i)-t 0 (i) In the above formula, Δt(i) represents the duration of the i-th third signal, t 0 (i) and t 1 (i) represents the initial time and the final time of the i-th third signal respectively; S36. Setting a duration threshold and selecting a third signal whose duration is greater than the duration threshold to generate a first fault signal; S4. Calculate the complexity of the first signal at a time corresponding to the first fault signal, and select the first fault signal corresponding to a time when the complexity is greater than a complexity threshold to generate a second fault signal; S5. Obtain the span duration range of the second fault signal and the blank time when the second fault signal does not exist, calculate the filling amount, set a failure rate function model, and then input the span duration range, blank time, and corresponding filling amount into the failure rate function model to obtain the failure rate; S6. Determine whether a wind turbine rotor fails based on whether the total failure rate of the second failure signals during the extreme weather period exceeds a failure rate threshold; If so, output a prompt message indicating that the wind turbine rotor is abnormal; If not, a prompt message indicating no abnormality is output.

2. The evaluation method according to claim 1, wherein: In step S2, the following steps are specifically included: S21. Establish a Kalman filter system model; S22. Setting a likelihood function for optimizing the model parameters of the Kalman filter system model, and obtaining the optimal parameters of the Kalman filter system model according to maximum likelihood estimation; the optimal parameters satisfy the following relationship: In the above formula, μ * represents the optimal parameters of the Kalman filter system model obtained when the likelihood function takes the minimum value, μ represents the model parameter set of the Kalman filter system model, represents the covariance matrix of the estimation error, represents the noise matrix, c1 represents the sampling length of the Kalman filter system model; S23, substituting the obtained optimal parameters into the Kalman filter system model, and filtering the vibration signal to obtain an optimal state matrix of the filtered vibration signal; S24. Obtain the amplitudes of different frequency signals at different times according to the optimal state matrix of the vibration signal to obtain a first signal; the expression of the first signal is: In the above formula, A i，k and The frequencies of the vibration signals at time k are The amplitude and initial phase of the signal component, x 2i，k Represents the 2i-th number in the optimal state matrix at time k.

3. The evaluation method according to claim 1, wherein: In step S4, the following steps are specifically included: S41, marking the time corresponding to the first fault signal as a reference time; S42. Set a number of amplitude intervals, calculate the probability of the first signal corresponding to the amplitude interval at any reference time, that is, the first amplitude probability, and normalize the first amplitude probability to obtain a second amplitude probability; S43. Calculate the complexity of the first signal at any moment according to the amplitude probability; S44 , identifying a moment when the complexity is greater than a complexity threshold, and selecting a first fault signal corresponding to the moment to obtain a second fault signal.

4. The evaluation method according to claim 3, wherein: In step S42, the following steps are specifically included: S421, setting a plurality of amplitude intervals, and making the amplitude value of each moment in the first signal correspond to each amplitude interval; S422. Calculate the probability of the first signal corresponding to the amplitude interval at any time, that is, the first amplitude probability. The calculation formula for the first amplitude probability is: In the above formula, It represents the probability of the first signal appearing in the corresponding amplitude interval at the rth reference moment, that is, the first amplitude probability, N r N represents the number of amplitude intervals corresponding to the first fault signal at the rth reference moment in the first signal, R represents the total number of amplitude intervals in the first signal; S423. Normalize the first amplitude probability to obtain a second amplitude probability of the first fault signal appearing in the amplitude interval corresponding to the first fault signal at any time. The calculation formula for the second amplitude probability is: In the above formula, represents the probability of the second amplitude occurring in the amplitude interval corresponding to the first signal at the rth moment, It represents the first amplitude probability of the first signal corresponding to the amplitude interval at the rth moment, and the total number of amplitude intervals on the first signal is q.

5. The evaluation method according to claim 1, wherein: In step S5, the following steps are specifically included: S51, obtaining a span duration range of the second fault signal from the first moment to the last moment; S52, marking moments that do not correspond to the second fault signal within the spanning time range as blank moments, and marking several adjacent blank moments as blank periods; S53. Setting a failure rate function model in which the failure rate changes with time. The failure rate function model satisfies the following relationship: In the above formula, f(t i ) indicates that at t i The output of the failure rate function model at time K i Indicates that at t i The slope of the failure rate function model at time K 0 represents the upper limit of the slope; S54. According to the failure rate function model, the function value at the time corresponding to the second fault signal is taken as a positive number, and the function value at the time corresponding to the blank period is taken as a negative number, and the total failure rate of the second fault signal within the span time range is calculated.

6. The evaluation method according to claim 5, characterized in that In step S52, the following steps are specifically included: S521, setting a time window; S522: Obtain a moment in the spanning time range that does not correspond to the second fault signal, and mark it as a blank moment; S523, traversing each blank moment in sequence through the time window, and determining whether the moments corresponding to both ends of the time window are the moments corresponding to the second fault signal; If yes, then mark the time corresponding to the time window as the time corresponding to the second fault signal; If not, proceed to the next step; S524: Mark several adjacent blank moments as a blank period.

7. The evaluation method according to claim 5, characterized in that In step S54, the following steps are specifically included: S541. Set several time intervals and set weight coefficients corresponding to the time intervals; S542. Calculate the filling amount of each blank moment and the second fault signal within the span duration range according to the weight coefficient. The calculation formula of the filling amount is: In the above formula, W i Indicates the filling amount of the i-th blank period or the second fault signal within the time range, t i1 and t im represents the first and last moments of the i-th blank period within the duration range, σ i Represents the time interval [t i1 , t im ]The corresponding weight coefficient, f(x) represents the failure rate function model; S543: Calculate the total failure rate of the second fault signal within the span time range according to the filling amount.

8. The evaluation method according to claim 3, wherein: In step S43, the complexity calculation formula is: In the above formula, FZ r represents the complexity of the first signal at the rth moment, is the probability of the second amplitude occurring in the amplitude interval corresponding to the rth moment at the Fth frequency. There are a total of F1 frequencies at the rth moment in the first signal.

9. The evaluation method according to claim 7, wherein: In step S543, the total failure rate is calculated as follows: In the above formula, P represents the total failure rate of the second fault signal within the span time range, W i and W j They represent the filling amount of the i-th second fault signal and the filling amount of the j-th blank period within the time range, T 1 represents the set of second fault signals, T 2 Represents a set of blank periods, t1 to t m That is, the first moment and the last moment of the span duration, σ0 represents the weight coefficient corresponding to the span duration, and f(x) represents the failure rate function model.

Images