Fault diagnosis processing method and device for main shaft of wind driven generator

Abstract

The invention provides a fault diagnosis processing method and device for a main shaft of a wind driven generator, belongs to the technical field of fault diagnosis processing of the wind driven generator, and solves the problems that the fault of the main shaft is not found timely and the diagnosis is inaccurate. The method comprises the following steps: acquiring a vibration signal characteristic value and current working condition data of a wind driven generator spindle; performing regression prediction processing according to the vibration signal characteristic value and the current working condition data to obtain a prediction error; determining an error vector according to the prediction error and a preset threshold value; obtaining fault type information and corresponding fault symptom information of a symptom library; vectorizing the fault type information and the fault symptom information to obtain a symptom vector; determining similarity according to the error vector and the symptom vector; determining the fault probability of the wind driven generator spindle according to the similarity and the fault symptom information; and outputting a fault diagnosis result according to the fault probability of the main shaft of the wind driven generator. According to the technical scheme, safe operation of the wind generating set can be guaranteed, and the operation and maintenance cost is reduced.

Description

Technical Field

[0001] This invention relates to the field of wind turbine generator fault diagnosis and treatment technology, and in particular to a method and apparatus for diagnosing and treating faults in the main shaft of a wind turbine generator. Background Technology

[0002] In wind power generation systems, the main shaft, as a key transmission component, plays a crucial role in converting the wind energy captured by the wind turbine into mechanical energy and transferring it to the generator. With the rapid development of the wind power industry, the single-unit capacity of wind turbine generators is constantly increasing, and the operating environment is becoming increasingly complex. The loads and stresses borne by the main shaft are also constantly increasing, which correspondingly increases the probability of main shaft failure.

[0003] Main shaft failure not only causes wind turbine generators to shut down and reduce power generation efficiency, but also triggers a series of chain reactions, such as increased wear on other components, affecting the stability of the entire drivetrain, and potentially causing serious equipment damage and economic losses. Statistics show that main shaft failure accounts for a significant proportion of wind turbine generator failures, with high repair costs and a complex and time-consuming repair process.

[0004] The vibration characteristics of the spindle change in real time with operating conditions such as rotational speed and wind speed, and have complex coupling relationships with other systems. It is difficult for general rule models to achieve early warning and monitoring. The main shaft is a major component of a large wind turbine. When the wind turbine alarms, it already faces the need for component replacement or has suffered a serious fault (shutdown for maintenance or severe damage to the wind turbine). Therefore, it is difficult to collect actual fault data, making it difficult to establish a diagnostic model. General diagnostic models are built based on experimental simulation data, which differs greatly from the actual unit conditions, resulting in inaccurate diagnosis of main shaft faults. Summary of the Invention

[0005] This invention provides a method and apparatus for diagnosing and handling faults in the main shaft of a wind turbine, which solves the problems of untimely detection and inaccurate diagnosis of main shaft faults in wind turbines, making it difficult to ensure the reliable operation of wind turbine units, resulting in low power generation efficiency and high operation and maintenance costs.

[0006] To solve the above-mentioned technical problems, the technical solution of the present invention is as follows: This invention provides a method for diagnosing and handling faults in the main shaft of a wind turbine generator, comprising: Obtain the vibration signal characteristic values ​​and current operating condition data of the wind turbine main shaft; The prediction error is obtained by performing regression prediction processing based on the vibration signal characteristic values ​​and the current working condition data. The error vector is determined based on the prediction error and the preset threshold. Obtain fault type information and corresponding fault symptom information from the symptom database; The fault type information and fault symptom information are vectorized to obtain symptom vectors; The similarity is determined based on the error vector and the symptom vector; Based on the similarity and the fault symptom information, the probability of wind turbine main shaft failure is determined; Based on the failure probability of the wind turbine main shaft, output the fault diagnosis result.

[0007] Optionally, the vibration signal characteristic values ​​of the wind turbine main shaft are obtained, including: Vibration signal data of the wind turbine are acquired by sensors installed on the wind turbine. The vibration signal data is processed to construct feature values, thereby obtaining the vibration signal feature values.

[0008] Optionally, the vibration signal data can be processed to construct feature values, including: Based on the acceleration sequence in the vibration signal data, feature value construction processing is performed on the vibration signal data to obtain the asymmetry feature value and the smoothness feature value of the vibration signal. Based on the frequency of the vibration signal data, feature value construction processing is performed on the vibration signal data to obtain the feature value of the degree of change of vibration signal components.

[0009] Optionally, regression prediction processing is performed based on the vibration signal feature values ​​and the current operating condition data to obtain the prediction error, including: The vibration signal feature value from the previous moment and the current operating condition data are input into a preset regression model to obtain the predicted value of the vibration signal feature value at the current moment. The prediction error is obtained by subtracting the predicted value of the vibration signal characteristic value at the current moment from the predicted value of the vibration signal characteristic value at the current moment.

[0010] Optionally, based on the prediction error and a preset threshold, an error vector is determined, including: Obtain preset alarm thresholds and preset fault thresholds; Based on the prediction error, the preset alarm threshold, and the preset fault threshold, an intermediate result is obtained; The intermediate results corresponding to each vibration signal feature value are vectorized to obtain the feature vector corresponding to each feature; The feature vectors of each feature are summarized and transposed to determine the error vector.

[0011] Optionally, the fault type information and fault symptom information are vectorized to obtain a symptom vector, including: Determine the number of fault symptoms for all fault types based on the fault symptom information; The symptom vector is obtained by taking the number of fault symptoms as the dimension of the symptom vector and using the symptom features corresponding to each fault type as elements of the symptom vector.

[0012] Optionally, determining the similarity based on the error vector and the symptom vector includes: according to Determine the similarity; in, To determine the similarity with the m-th symptom vector, For the error vector, This is the symptom vector.

[0013] Optionally, the probability of a wind turbine main shaft failure is determined based on the similarity and the fault symptom information, including: according to Determine the probability of failure; in, Let be the probability of the m-th symptom vector corresponding to the error vector. Let be the similarity of the m-th fault corresponding to the error vector. This represents the number of symptoms corresponding to the m-th fault. Each error yields the probability of m faults. .

[0014] This invention also provides a wind turbine main shaft fault diagnosis and processing device, comprising: The acquisition module is used to acquire the vibration signal characteristic values ​​and current operating condition data of the wind turbine main shaft; and to acquire fault type information and corresponding fault symptom information from the symptom database. The processing module is used to perform regression prediction processing on the vibration signal feature values ​​and the current operating condition data to obtain a prediction error; determine an error vector based on the prediction error and a preset threshold; vectorize the fault type information and fault symptom information to obtain a symptom vector; determine the similarity based on the error vector and the symptom vector; and determine the probability of a wind turbine main shaft failure based on the similarity and the fault symptom information. The output module is used to output fault diagnosis results based on the fault probability of the wind turbine main shaft.

[0015] The technical solution of the present invention has at least the following effects: The above-mentioned solution of the present invention obtains the vibration signal feature values ​​and current operating condition data of the wind turbine main shaft; performs regression prediction processing based on the vibration signal feature values ​​and the current operating condition data to obtain a prediction error; determines an error vector based on the prediction error and a preset threshold; obtains fault type information and corresponding fault symptom information from a symptom database; vectorizes the fault type information and fault symptom information to obtain a symptom vector; determines the similarity between the error vector and the symptom vector; determines the fault probability of the wind turbine main shaft based on the similarity and the fault symptom information; and outputs a fault diagnosis result based on the fault probability of the wind turbine main shaft. The technical solution of the present invention can ensure the safe operation of wind turbine units and extend the service life of equipment by preventing major accidents of the wind turbine main shaft, while also reducing operation and maintenance costs and improving power generation efficiency by reducing downtime. Attached Figure Description

[0016] Figure 1 This is a flowchart of the wind turbine main shaft fault diagnosis and processing method provided in the embodiments of the present invention; Figure 2 This is a flowchart of the module of the wind turbine main shaft fault diagnosis and processing method provided in the embodiment of the present invention; Figure 3 This is a structural diagram of the wind turbine main shaft fault diagnosis and processing device provided in an embodiment of the present invention. Detailed Implementation

[0017] Exemplary embodiments of the invention will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the invention are shown in the drawings, it should be understood that the invention may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this invention will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art.

[0018] like Figure 1 and Figure 2 As shown, an embodiment of the present invention proposes a method for diagnosing and processing faults in the main shaft of a wind turbine generator, comprising: Step 11: Obtain the vibration signal characteristic values ​​and current operating condition data of the wind turbine main shaft; Step 12: Perform regression prediction processing based on the vibration signal feature values ​​and the current working condition data to obtain the prediction error; Step 13: Determine the error vector based on the prediction error and the preset threshold; Step 14: Obtain the fault type information and corresponding fault symptom information from the symptom database; Step 15: Vectorize the fault type information and fault symptom information to obtain symptom vectors; Step 16: Determine the similarity based on the error vector and the symptom vector; Step 17: Determine the probability of wind turbine main shaft failure based on the similarity and the fault symptom information; Step 18: Output the fault diagnosis result based on the fault probability of the wind turbine main shaft.

[0019] In this embodiment, vibration sensors installed on the wind turbine collect vibration signal data of the main shaft in real time during operation. The collected vibration signal data contains important information about the wind turbine's operating status, such as vibration frequency, amplitude, phase, and acceleration sequence. Subsequently, general calculation methods or signal processing techniques, including Fourier transform and function construction, are used to process the vibration signal data and extract vibration signal feature values ​​that reflect the operating status of the wind turbine's main shaft. Simultaneously, it is also necessary to collect current wind turbine operating condition data, including load, speed, power, and wind speed, for subsequent operations.

[0020] By using regression analysis and combining historical data with the currently collected vibration signal characteristic values ​​and operating condition data, a prediction model is constructed. This model can predict the current spindle vibration signal characteristic value, i.e., the predicted value, given the current operating condition and the vibration signal characteristic value of the previous moment. The prediction error is calculated by subtracting the predicted value output by the prediction model from the currently collected vibration signal characteristic value. The prediction error reflects the degree of deviation between the current spindle vibration state and the expected normal state.

[0021] The prediction error is compared with a preset threshold, which is set based on historical data and experience. This includes a preset alarm threshold and a preset fault threshold. Different intermediate results are obtained based on different prediction error values ​​within different preset threshold ranges. The intermediate results are then processed to form an error vector.

[0022] The symptom database is a pre-built database that stores various types of faults that may occur on the main shaft of a wind turbine generator, as well as the corresponding symptom information for each type of fault. The fault type information describes the type and nature of the fault, while the symptom information describes the various phenomena and manifestations that will occur when the fault occurs. In order to provide a basis for subsequent analysis and diagnosis, this information needs to be obtained from the symptom database.

[0023] To facilitate computation, the fault type information and fault symptom information need to be vectorized. The dimension of the symptom vector is determined based on the fault symptom information, and the symptom features corresponding to each fault type are used as elements of the symptom vector to obtain a numerical vector, namely the symptom vector.

[0024] Similarity measures the degree of closeness between the error vector and the symptom vector. It can be calculated using methods such as cosine similarity and Euclidean distance. By comparing the similarity between the error vector and the symptom vectors corresponding to various fault types, the abnormal state of the current spindle can be determined. The higher the similarity, the more the current fault state of the spindle matches the current fault type symptom, and therefore the greater the probability of that fault type occurring.

[0025] By combining the similarity calculation results with the number of symptoms corresponding to the fault type, the probability of various faults occurring on the main shaft of the wind turbine can be further determined.

[0026] Based on the different probabilities of wind turbine main shaft failures, different fault diagnosis results are output. The fault diagnosis result information includes whether the fault type exists in the expert database, the specific fault type in the corresponding expert database, and the fault probability of the fault type. The fault diagnosis result information provides maintenance personnel with maintenance references, enabling them to understand the operating status of the main shaft in a timely manner and take corresponding maintenance measures to ensure the safe and stable operation of the wind turbine.

[0027] This technical solution performs feature value construction processing on vibration signal data to obtain real vibration signal feature value information, providing practical data support for the establishment of diagnostic models. This makes the model more closely resemble the actual operation of wind turbine generators, providing reliable prediction results for wind turbine generator main shaft failures, thereby extending equipment service life and ensuring stable equipment operation.

[0028] In an optional embodiment of the present invention, step 11 may include: Step 111: Obtain vibration signal data of the wind turbine through sensors installed on the wind turbine. Step 112: Perform feature value construction processing on the vibration signal data to obtain vibration signal feature values; Step 113: Obtain current operating condition data through sensors installed on the wind turbine; The process of constructing feature values ​​for vibration signal data includes: constructing feature values ​​for vibration signal data based on the acceleration sequence in the vibration signal data to obtain asymmetry feature values ​​and smoothness feature values; and constructing feature values ​​for vibration signal data based on the frequency of the vibration signal data to obtain component variation feature values.

[0029] In this embodiment, the sensor is installed in the bearing housing of the spindle, the connection between the spindle and the gearbox, etc., to ensure that it can accurately reflect the vibration state of the spindle. During installation, the fixing method, orientation and contact quality with the measured surface of the sensor are considered to reduce measurement errors. The collected data is stored in a database or file for subsequent processing and analysis.

[0030] Feature values ​​are obtained by using common feature calculation methods in the time or frequency domain, or feature values ​​are constructed by using the symptoms of corresponding faults. The feature values ​​obtained by using common feature calculation methods in the time or frequency domain include at least one of mean, variance, effective value, peak-to-peak value, skewness, kurtosis, and mean square frequency. The feature values ​​constructed by using the symptoms of corresponding faults include at least one of asymmetry feature value, smoothness feature value, and component variability feature value.

[0031] Current operating condition data is acquired by sensors installed on the wind turbine, including at least one of rotational speed, power, wind speed, and load.

[0032] In an optional embodiment of the present invention, step 112 involves: performing feature value construction processing on the vibration signal data based on the acceleration sequence in the vibration signal data to obtain the asymmetric feature values ​​of the vibration signal, including: Step 1121, according to Obtain the signal asymmetry eigenvalues; In an optional embodiment of the present invention, step 112 involves: performing feature value construction processing on the vibration signal data based on the acceleration sequence in the vibration signal data to obtain the smoothness feature value of the vibration signal, including: Step 1122, according to Obtain signal smoothness feature values; in, The acceleration sequence of the vibration signal. The average value of the acceleration sequence of the vibration signal. is the length of the vibration acceleration sequence.

[0033] In an optional embodiment of the present invention, step 112 involves: performing feature value construction processing on the vibration signal data based on the frequency of the vibration signal data to obtain feature values ​​of the degree of change of vibration signal components, including: Step 1123: Perform Fourier transform on the vibration signal to obtain the amplitude of all frequencies, and then remove the top three main frequencies to obtain the amplitude of the remaining frequencies. Step 1124, according to Obtain characteristic values ​​of the degree of change in vibration signal components; in, It is the sum of the amplitudes of all frequencies in the vibration signal data. It is the sum of the amplitudes of the top 3 vibration signal data.

[0034] In this embodiment, by performing feature analysis on the vibration signal of the wind turbine main shaft, the vibration characteristics change in real time with operating conditions such as rotational speed and wind speed. This can reflect the coupling relationship between vibration and other systems, providing more accurate feature data in subsequent fault prediction, thereby improving the accuracy and real-time performance of subsequent fault prediction.

[0035] In an optional embodiment of the present invention, step 12 may include: Step 121: Input the vibration signal feature value of the previous moment and the current working condition data into a preset regression model to obtain the predicted value of the vibration signal feature value at the current moment; Step 122: Subtract the predicted value of the vibration signal characteristic value at the current time from the predicted value of the vibration signal characteristic value at the current time to obtain the prediction error.

[0036] In this embodiment, the process of using a regression prediction model to predict the characteristic values ​​of vibration signals and evaluating the operating status or potential faults of the spindle by calculating the error between the predicted value and the actual value mainly includes two core steps: predictive value generation and prediction error calculation.

[0037] The prediction generation process involves inputting the vibration signal feature value from the previous moment and the operating condition data from the current moment into a preset regression model. The vibration signal feature value from the previous moment reflects the operating state of the main shaft at the previous moment and can provide a certain prediction basis for the vibration signal feature value at the current moment. Therefore, in this model, the vibration signal feature value from the previous moment is mandatory data. according to The vibration signal feature value from the previous moment and the current operating condition data are processed to obtain the predicted value of the feature value at the current moment; in, For a well-trained regression model, The feature value of the previous time step. This is the current operating condition data. This is the predicted value of the feature value at the current time.

[0038] The prediction error calculation process is based on - The prediction error is obtained, where, The prediction error is... The current feature value is denoted as ; the prediction error reflects the degree of deviation between the predicted value and the actual feature value, and is an important indicator for evaluating the prediction performance of the regression model.

[0039] In an optional embodiment of the present invention, step 13 may include: Step 131: Obtain the preset alarm threshold and preset fault threshold; Step 132: Obtain intermediate results based on the prediction error, preset alarm threshold, and preset fault threshold; Step 133: Vectorize the intermediate results corresponding to each vibration signal feature value to obtain the feature vector corresponding to each feature; Step 134: Summarize and transpose the feature vectors of each feature to determine the error vector.

[0040] In this embodiment, when the prediction error exceeds a preset threshold, it indicates that the spindle's operating state has changed abnormally, requiring further fault diagnosis and processing. The preset threshold includes a preset alarm threshold. and preset fault thresholds The preset alarm threshold is less than the preset fault threshold, that is This ensures that alarms precede fault diagnosis, providing sufficient response time.

[0041] In step 132, the current abnormal state is classified according to the prediction error value. At this point, the situation is normal, so x is assigned the value 0 to indicate that there is no abnormality; when Currently, the system is in an alarm state; assign a value to x. (1≥x≥0), indicating that the degree of anomaly increases with increasing error; when At this point, the system is in a fault state, so x is assigned the value 1, indicating that the fault determination threshold has been reached. Specifically, based on...

[0042] Intermediate results are obtained to quantify the degree of anomaly, whereby... The preset alarm threshold, The preset fault threshold, The prediction error is... This is an intermediate result; In step 133, according to ; ;...

[0043] The intermediate results are vectorized to obtain the feature vector corresponding to each feature. Each feature vector has a non-zero value only at the corresponding position, which facilitates subsequent matrix operations. Here, n is the number of feature vectors. Let x1 be the first element of the feature vector of the first feature. Let x2 be the eigenvector of the second feature, and x2 be the second element of the eigenvector of the second feature... Let x be the eigenvector of the nth feature. n The nth element of the eigenvector of the nth feature; In step 134, the feature vectors of each feature are summarized and transposed, according to...

[0044] The resulting error vector is used to integrate the anomaly levels of all features, serving as input for the similarity calculation process. This is the error vector.

[0045] By using the intermediate results and the error vector, the abstract abnormal state is transformed into a quantifiable value, which facilitates analysis and comparison. The alarm threshold and the fault threshold are designed in a hierarchical manner to achieve a progressive response from early warning to fault confirmation, providing a reliable basis for subsequent similarity calculation.

[0046] In an optional embodiment of the present invention, step 14, obtaining the fault type information and corresponding fault symptom information from the symptom database, includes: Step 141: Establish a symptom database based on expert experience, and store fault type information and fault symptom information; Step 142: Extract the required information from the symptom database.

[0047] In this embodiment, the symptom database is a structured knowledge base used to store fault type information of the wind turbine main shaft and its corresponding fault symptom information. The information filling of the symptom database mainly relies on past expert experience and historical maintenance data of wind turbines to provide a reference for fault diagnosis. The symptom database can systematize expert experience and avoid knowledge loss due to personnel turnover. As expert experience and maintenance data grow, the information in the symptom database will be supplemented synchronously, and the accuracy of the information in the symptom database will be verified through feedback from actual maintenance results, thereby correcting and improving the data.

[0048] Fault type information describes the specific nature of the fault, while fault symptom information describes the specific manifestations of the fault. Each fault type corresponds to one or more fault symptom information. Partial information from the symptom database is shown in Table 1.

[0049] Table 1. Partial Information from the Symptom Database

[0050] In an optional embodiment of the present invention, step 15, which involves vectorizing the fault type information and fault symptom information to obtain a symptom vector, may include: Step 151: Determine the number of fault symptoms for all fault types based on the fault symptom information; Step 152: Using the number of fault symptoms as the dimension of the symptom vector, the symptom features corresponding to each fault type are used as elements of the symptom vector to obtain the symptom vector.

[0051] In this embodiment, fault type information and fault symptom information are transformed into structured symptom vectors, which can convert discrete text descriptions into numerical vectors, making it easier for the program to calculate and process them, and providing standardized input for subsequent similarity calculations.

[0052] The entire list of fault symptoms in the symptom database is aggregated and deduplicated to obtain the dimension of the symptom vector and the corresponding symptom. Specifically, all fault types in the symptom database are traversed, their corresponding fault symptom information is extracted, and aggregated into a set. Duplicate symptom information is removed to ensure that each symptom appears only once, resulting in a unique list of fault symptoms. The length of this list is the dimension of the symptom vector. For each fault type, it is traversed, and for each fault type, it is checked whether its corresponding symptom exists in the entire list of fault symptoms. If the symptom exists, the value of the corresponding element in the symptom vector is set to 1; if the symptom does not exist, the value of the corresponding element is set to 0. This ensures that each fault type yields a symptom vector with the same dimension.

[0053] The expert symptom database is vectorized. Each fault has one or more different symptom manifestations. For example, if there are 10 characteristic symptoms that can reflect all fault types, then the dimension of the expert symptom vector is 10-dimensional. Each fault is iterated over; if the fault is reflected in the current symptom, it is set to 1; otherwise, it is set to 0. In this way, each fault will generate a symptom vector of the same dimension. Examples of symptom vectorization are shown in Table 2.

[0054] Table 2 Examples of Symptom Vectorization

[0055] In an optional embodiment of the present invention, step 16, determining the similarity based on the error vector and the symptom vector, may include: Step 161, according to Determine the similarity, where, To determine the similarity with the m-th symptom vector, For the error vector, This is the symptom vector.

[0056] In this embodiment, the similarity between the error vector and the symptom vector is calculated to quantify the degree of matching between the error vector and the symptom of each fault type, providing a quantitative basis for the subsequent fault probability calculation process.

[0057] The error vector and symptom vector are input into a formula to calculate the similarity data. The smaller the similarity value, the closer the error vector is to the symptom vector. mThe higher the similarity between the symptom vectors, the more similar the results. Since the error vector and symptom vectors have a one-to-many relationship, the number of similarity scores obtained is the same as the number of symptom vectors. The similarity between the error vector and the first symptom vector. The similarity between the error vector and the second symptom vector... Let be the similarity between the error vector and the m-th symptom vector.

[0058] In an optional embodiment of the present invention, step 17, determining the probability of a wind turbine main shaft failure based on the similarity and the fault symptom information, may include: Step 171, according to Determine the failure probability for each type of fault, where, Let be the probability of the m-th symptom vector corresponding to the error vector. Let be the similarity of the m-th fault corresponding to the error vector. This represents the number of symptoms corresponding to the m-th fault. Step 172: For each error, obtain the probability of m faults. .

[0059] In this embodiment, the original similarity value is converted into a fault probability by combining the similarity with the number of symptoms. Since the number of symptoms for each fault is different, directly using the similarity would lead to unfair comparisons. Normalization can eliminate this effect. Therefore, the similarity is normalized according to the corresponding number of symptoms in the symptom database. The fault probability is obtained by comparing the similarity with the fault symptom information in the expert symptom database. The higher the fault probability value, the higher the matching degree between the fault type and the current error vector, and the influence of the number of symptoms has been eliminated.

[0060] In an optional embodiment of the present invention, step 18, outputting a fault diagnosis result based on the fault probability of the wind turbine main shaft, may include: Step 181: Output different wind turbine fault diagnosis results based on different fault probabilities.

[0061] In this embodiment, if the probability of all faults is less than 0.5, the output is directly: "Spindle vibration abnormal, fault type not in the expert database"; if only one fault has a probability greater than 0.5, the output is: "Spindle vibration abnormal, fault type is {fault type corresponding to greater than 0.5}"; if multiple faults have probabilities greater than 0.5, in order to make the sum of all output probability values ​​equal to 1, the final output result needs to be calculated based on all probability values ​​greater than 0.5: according to Perform the final probability calculation, where, For faults with a probability greater than 0.5, Let be the final probability of fault 1. The final probability of fault 2... Let n be the final probability of fault n; the final fault diagnosis result when there are multiple probabilities greater than 0.5 will be output: "Abnormal spindle vibration, the probability of {fault 1} occurring is { The probability of {fault 2} occurring is { The probability of fault n occurring is { }".

[0062] A specific embodiment of the wind turbine main shaft fault diagnosis and processing method provided in this invention is as follows: Step 1: Obtain the vibration signal characteristic values ​​and current operating condition data of the wind turbine main shaft.

[0063] Specifically, vibration signal data of the wind turbine is acquired through sensors installed on the wind turbine; characteristic values ​​are obtained through common feature calculation methods in the time or frequency domain, or characteristic values ​​are constructed through the symptoms of corresponding faults. The characteristic values ​​obtained through common feature calculation methods in the time or frequency domain include at least one of mean, variance, RMS value, peak-to-peak value, skewness, kurtosis, and mean square frequency. The characteristic values ​​constructed through the symptoms of corresponding faults include at least one of asymmetry characteristic value, smoothness characteristic value, and component variability characteristic value; current operating condition data is acquired through sensors installed on the wind turbine, and the current operating condition data includes at least one of speed, power, wind speed, and load.

[0064] Based on the acceleration sequence in the vibration signal data, eigenvalue construction processing is performed on the vibration signal data to obtain the asymmetric eigenvalues ​​of the vibration signal, including: based on Obtain signal asymmetry feature values; based on the acceleration sequence in the vibration signal data, perform feature value construction processing on the vibration signal data to obtain the smoothness feature values ​​of the vibration signal, including: based on Obtain signal smoothness feature values; where, The acceleration sequence of the vibration signal. The average value of the acceleration sequence of the vibration signal. The length of the vibration acceleration sequence. Based on the frequency of the vibration signal data, feature value construction is performed on the vibration signal data to obtain the characteristic values ​​of the degree of change of vibration signal components. This includes: performing a Fourier transform on the vibration signal to obtain the amplitude of all frequencies, and then removing the top three dominant frequencies to obtain the amplitude of the remaining frequencies; based on... Obtain the characteristic values ​​of the degree of change in vibration signal components; where, It is the sum of the amplitudes of all frequencies in the vibration signal data. It is the sum of the amplitudes of the top 3 vibration signal data.

[0065] Step 2: Perform regression prediction processing based on the vibration signal feature values ​​and the current working condition data to obtain the prediction error.

[0066] Specifically, the vibration signal feature values ​​from the previous moment and the current operating condition data are input into a preset regression model to obtain the predicted values ​​of the vibration signal feature values ​​at the current moment. The feature value from the previous moment and the current operating condition data are processed to obtain the predicted value of the feature value at the current moment, wherein, For a well-trained regression model, The feature value of the previous time step. This is the current operating condition data. The predicted value of the characteristic value at the current moment; the difference between the predicted value of the vibration signal characteristic value at the current moment and the current vibration signal characteristic value, according to... - The prediction error is obtained, where, The prediction error is... This represents the feature value at the current moment.

[0067] Step 3: Determine the error vector based on the prediction error and the preset threshold.

[0068] Specifically, a preset alarm threshold and a preset fault threshold are obtained; intermediate results are obtained based on the prediction error, the preset alarm threshold, and the preset fault threshold; the intermediate results corresponding to each vibration signal feature value are vectorized to obtain the feature vector corresponding to each feature; the feature vectors of each feature are summarized and transposed to determine the error vector.

[0069] The preset alarm threshold is less than the preset fault threshold, according to

[0070] The intermediate results were obtained, among which, The preset alarm threshold, The preset fault threshold, The prediction error is... This is an intermediate result; according to ; ;...

[0071] The intermediate results are vectorized to obtain the feature vector corresponding to each feature, where n is the number of feature vectors. Let x1 be the first element of the feature vector of the first feature. Let x2 be the eigenvector of the second feature, and x2 be the second element of the eigenvector of the second feature... Let x be the eigenvector of the nth feature. n The nth element of the eigenvector of the nth feature; Summarize and transpose the feature vectors of each feature, according to

[0072] The error vector is obtained, where, This is the error vector.

[0073] Step 4: Obtain the fault type information and corresponding fault symptom information from the symptom database.

[0074] Specifically, a symptom database is established based on expert experience, storing fault type information and fault symptom information; the required information is then extracted from the database. The database is primarily populated using past experience in wind turbine main shaft maintenance, and the information is updated in tandem as expert experience grows. Fault type information describes the specific nature of the fault, while fault symptom information describes the specific manifestations of the fault. Each fault type corresponds to one or more fault symptom information.

[0075] Step 5: Vectorize the fault type information and fault symptom information to obtain symptom vectors.

[0076] Specifically, the number of fault symptoms for all fault types is determined based on the fault symptom information; the symptom vector is obtained by taking the number of fault symptoms as the dimension of the symptom vector and using the symptom features corresponding to each fault type as elements of the symptom vector.

[0077] In this embodiment, all fault symptom information in the symptom database is summarized and deduplicated to obtain the dimension of the symptom vector and the corresponding symptom. Each fault type is iterated over; if the fault type is reflected in the current symptom, the element value corresponding to the symptom vector is set to 1; otherwise, it is set to 0. This ensures that each fault yields a symptom vector of the same dimension. .

[0078] Step 6: Determine the similarity based on the error vector and the symptom vector.

[0079] Specifically, according to Determine the similarity, where, To determine the similarity with the m-th symptom vector, For the error vector, The error vector and the symptom vector are input into a formula for calculation to obtain similarity data. Since the error vector and the symptom vector have a one-to-many relationship, the number of similarity scores obtained is the same as the number of symptom vectors. The similarity between the error vector and the first symptom vector. The similarity between the error vector and the second symptom vector... Let be the similarity between the error vector and the m-th symptom vector.

[0080] Step 7: Determine the probability of wind turbine main shaft failure based on the similarity and the fault symptom information.

[0081] Specifically, the similarity is normalized based on the number of corresponding symptoms in the expert symptom database. Since the number of symptoms is different for each fault, the obtained similarity needs to be normalized to convert it into a fault probability based on the number of symptoms. Determine the failure probability for each type of fault, where, Let be the probability of the m-th symptom vector corresponding to the error vector. Let be the similarity of the m-th fault corresponding to the error vector. Let m be the number of symptom markers corresponding to the m-th fault; and let m be the probability of each error. .

[0082] Step 8: Output the fault diagnosis result based on the fault probability of the wind turbine main shaft.

[0083] Specifically, different wind turbine fault diagnosis results are output based on different fault probabilities. If all probabilities are less than 0.5, the output is: Main shaft vibration abnormal, fault type not in the expert database. If only one probability is greater than 0.5, the output is: Main shaft vibration abnormal, fault type {fault type corresponding to greater than 0.5}. If multiple probabilities are greater than 0.5, in order to ensure that the sum of all output probability values ​​is 1, the final output result needs to be calculated based on all probability values ​​greater than 0.5. Perform the final probability calculation, where, For faults with a probability greater than 0.5, Let be the final probability of fault 1. The final probability of fault 2... Let n be the final probability of fault n; output the final fault diagnosis result when there are multiple probabilities greater than 0.5: abnormal spindle vibration, the probability of {fault 1} occurring is { The probability of {fault 2} occurring is { The probability of {fault n} occurring is { }

[0084] The wind turbine main shaft fault diagnosis and processing method proposed in the above embodiments of the present invention, by vectorizing error information and symptom information, obtains the fault type and fault probability of wind turbine main shaft faults, which can detect potential faults in advance and reduce the occurrence of catastrophic accidents; improve power generation efficiency and operation and maintenance efficiency through precise maintenance and planned maintenance; avoid the expansion and development of faults and reduce the economic losses of downtime maintenance.

[0085] like Figure 3 As shown, this embodiment of the invention also provides a wind turbine main shaft fault diagnosis and processing device 60, comprising: The acquisition module 61 is used to acquire the vibration signal characteristic values ​​and current operating condition data of the wind turbine main shaft; and to acquire the fault type information and corresponding fault symptom information from the symptom database. Processing module 62 is used to perform regression prediction processing on the vibration signal feature values ​​and the current working condition data to obtain a prediction error; determine an error vector based on the prediction error and a preset threshold; vectorize the fault type information and fault symptom information to obtain a symptom vector; determine the similarity based on the error vector and the symptom vector; and determine the probability of a wind turbine main shaft failure based on the similarity and the fault symptom information. Output module 63 is used to output fault diagnosis results based on the fault probability of the wind turbine main shaft.

[0086] Optionally, the acquisition module 61 is specifically used to: acquire vibration signal data of the wind turbine through sensors installed on the wind turbine; and perform feature value construction processing on the vibration signal data to obtain vibration signal feature values.

[0087] Optionally, the processing module 62 is specifically used to: perform feature value construction processing on the vibration signal data based on the acceleration sequence in the vibration signal data to obtain the asymmetry feature value and the smoothness feature value of the vibration signal; Based on the frequency of the vibration signal data, feature value construction processing is performed on the vibration signal data to obtain the feature value of the degree of change of vibration signal components.

[0088] Optionally, regression prediction processing is performed based on the vibration signal feature values ​​and the current operating condition data to obtain the prediction error, including: The vibration signal feature value from the previous moment and the current operating condition data are input into a preset regression model to obtain the predicted value of the vibration signal feature value at the current moment. The prediction error is obtained by subtracting the predicted value of the vibration signal characteristic value at the current moment from the predicted value of the vibration signal characteristic value at the current moment.

[0089] Optionally, based on the prediction error and a preset threshold, an error vector is determined, including: Obtain preset alarm thresholds and preset fault thresholds; Based on the prediction error, the preset alarm threshold, and the preset fault threshold, an intermediate result is obtained; The intermediate results corresponding to each vibration signal feature value are vectorized to obtain the feature vector corresponding to each feature; The feature vectors of each feature are summarized and transposed to determine the error vector.

[0090] Optionally, the processing module 62 is also specifically used for: The fault type information and fault symptom information are vectorized to obtain symptom vectors, including: Determine the number of fault symptoms for all fault types based on the fault symptom information; The symptom vector is obtained by taking the number of fault symptoms as the dimension of the symptom vector and using the symptom features corresponding to each fault type as elements of the symptom vector.

[0091] Optionally, determining the similarity based on the error vector and the symptom vector includes: according to Determine the similarity; in, To determine the similarity with the m-th symptom vector, For the error vector, This is the symptom vector.

[0092] Optionally, the probability of a wind turbine main shaft failure is determined based on the similarity and the fault symptom information, including: according to Determine the probability of failure; in, Let be the probability of the m-th symptom vector corresponding to the error vector. Let be the similarity of the m-th fault corresponding to the error vector. This represents the number of symptoms corresponding to the m-th fault. Each error yields the probability of m faults. ; If there are multiple probabilities greater than 0.5, in order to ensure that the sum of all probability values ​​in the output is 1, the final output result needs to be calculated based on all probability values ​​greater than 0.5. Perform the final probability calculation.

[0093] Optionally, output module 63 is specifically used for: Different wind turbine fault diagnosis results are output based on different fault probabilities. If all probabilities are less than 0.5, the output is: Main shaft vibration abnormal, fault type not in the expert database. If only one probability is greater than 0.5, the output is: Main shaft vibration abnormal, fault type {fault type corresponding to greater than 0.5}. If multiple probabilities are greater than 0.5, in order to ensure that the sum of all output probability values ​​is 1, the final output result needs to be calculated based on all probability values ​​greater than 0.5. Perform the final probability calculation, where, For faults with a probability greater than 0.5, Let be the final probability of fault 1. The final probability of fault 2... Let n be the final probability of fault n; output the final fault diagnosis result when there are multiple probabilities greater than 0.5: abnormal spindle vibration, the probability of {fault 1} occurring is { The probability of {fault 2} occurring is { The probability of {fault n} occurring is { .

[0094] It should be noted that this device is a device corresponding to the above method. All implementation methods in the above method embodiments are applicable to this embodiment and can achieve the same technical effect.

[0095] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0096] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0097] In the embodiments provided by this invention, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0098] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0099] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0100] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, ROM, RAM, magnetic disks, or optical disks.

[0101] Furthermore, it should be noted that in the apparatus and method of the present invention, it is obvious that the components or steps can be decomposed and / or recombined. These decompositions and / or recombinations should be considered equivalent solutions of the present invention. Moreover, the steps performing the above series of processes can naturally be executed in the order described, but are not necessarily required to be executed in chronological order; some steps can be executed in parallel or independently of each other. Those skilled in the art will understand that all or any step or component of the method and apparatus of the present invention can be implemented in any computing device (including processors, storage media, etc.) or network of computing devices, in hardware, firmware, software, or a combination thereof. This is something that those skilled in the art can achieve by using their basic programming skills after reading the description of the present invention.

[0102] Therefore, the object of the present invention can also be achieved by running a program or a set of programs on any computing device. The computing device can be a known general-purpose device. Therefore, the object of the present invention can also be achieved simply by providing a program product containing program code for implementing the method or apparatus. That is, such a program product also constitutes the present invention, and the storage medium storing such a program product also constitutes the present invention. Obviously, the storage medium can be any known storage medium or any storage medium developed in the future. It should also be noted that in the apparatus and method of the present invention, it is obvious that the components or steps can be decomposed and / or recombined. These decompositions and / or recombinations should be considered equivalent to the present invention. Furthermore, the steps for performing the above series of processes can naturally be performed in the order described, but are not necessarily required to be performed in chronological order. Some steps can be performed in parallel or independently of each other.

[0103] The above are preferred embodiments of the present invention. It should be noted that, for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A wind turbine main shaft failure diagnosis processing method characterized by, The method comprises the following steps: obtaining vibration signal characteristic values and current working condition data of a wind turbine main shaft; performing regression prediction processing on the vibration signal characteristic values and the current working condition data to obtain a prediction error; determining an error vector according to the prediction error and a preset threshold value; obtaining fault type information and corresponding fault symptom information of a symptom library; vectorizing the fault type information and the fault symptom information to obtain a symptom vector; determining a similarity according to the error vector and the symptom vector; determining a wind turbine main shaft fault probability according to the similarity and the fault symptom information; outputting a fault diagnosis result according to the wind turbine main shaft fault probability.

2. The wind generator main shaft failure diagnosis processing method according to claim 1, characterized by, The method comprises the following steps: obtaining vibration signal characteristic values of a wind turbine main shaft comprises the following steps: obtaining vibration signal data of a wind turbine through a sensor installed on the wind turbine; 3. The wind generator main shaft failure diagnosis processing method according to claim 2, characterized by, performing characteristic value construction processing on the vibration signal data to obtain vibration signal characteristic values. The method comprises the following steps: performing characteristic value construction processing on the vibration signal data to obtain vibration signal characteristic values comprises the following steps:

4. The wind generator main shaft failure diagnosis processing method according to claim 1, characterized by, performing characteristic value construction processing on the vibration signal data according to an acceleration sequence in the vibration signal data to obtain an asymmetry characteristic value of the vibration signal and a smoothness characteristic value of the vibration signal; performing characteristic value construction processing on the vibration signal data according to a frequency of the vibration signal to obtain a vibration signal component variation degree characteristic value. The method comprises the following steps:

5. The wind generator main shaft failure diagnosis processing method according to claim 1, characterized by, performing regression prediction processing on the vibration signal characteristic values and the current working condition data to obtain a prediction error comprises the following steps: inputting the vibration signal characteristic values of the last moment and the current working condition data into a preset regression model to obtain a predicted value of the vibration signal characteristic values of the current moment; subtracting the predicted value of the vibration signal characteristic values of the current moment from the vibration signal characteristic values of the current moment to obtain a prediction error. The method comprises the following steps: determining an error vector according to the prediction error and a preset threshold value comprises the following steps:

6. The wind generator main shaft failure diagnosis processing method according to claim 1, characterized by, obtaining a preset alarm threshold value and a preset fault threshold value; obtaining an intermediate result according to the prediction error, the preset alarm threshold value and the preset fault threshold value; vectorizing an intermediate result corresponding to each vibration signal characteristic value to obtain a characteristic vector corresponding to each characteristic; 7. The wind generator main shaft failure diagnosis processing method according to claim 1, characterized by, summarizing and transposing the characteristic vectors of each characteristic to determine an error vector. According to determining the similarity; wherein, is a similarity to the mth symptom vector, is an error vector, is a symptom vector.

8. The wind generator main shaft failure diagnosis processing method according to claim 1, characterized by, The method comprises the following steps: According to determining a probability of failure; wherein, is a probability of the mth symptom vector corresponding to the error vector, is a similarity of the mth fault corresponding to the error vector, is a number of symptom items corresponding to the mth fault. The probability of m failures per error .

9. A wind turbine main shaft failure diagnosis processing apparatus characterized by comprising: vectorizing the fault type information and the fault symptom information to obtain a symptom vector comprises the following steps: determining the number of fault symptoms of all fault types according to the fault symptom information; taking the number of fault symptoms as the dimension of the symptom vector, taking the symptom characteristics corresponding to each fault type as the elements of the symptom vector, and obtaining a symptom vector. The method comprises the following steps: determining a similarity according to the error vector and the symptom vector comprises the following steps: determining a wind turbine main shaft fault probability according to the similarity and the fault symptom information comprises the following steps: The method comprises the following steps: an obtaining module is configured to obtain vibration signal characteristic values and current working condition data of a wind turbine main shaft; a symptom library is configured to obtain fault type information and corresponding fault symptom information; The processing module is configured to perform regression prediction processing on the vibration signal characteristic value and the current working condition data to obtain a prediction error, determine an error vector according to the prediction error and a preset threshold, perform vectorization processing on the fault type information and the fault symptom information to obtain a symptom vector, determine a similarity according to the error vector and the symptom vector, and determine a wind turbine main shaft fault probability according to the similarity and the fault symptom information. The output module is configured to output a fault diagnosis result according to the wind turbine main shaft fault probability.

10. A computer readable storage medium characterized by, The computer is caused to perform the method according to any one of claims 1 to 8 when the instructions are executed on the computer.

Images