CN110688617A - Fan vibration abnormity detection method and device - Google Patents
Fan vibration abnormity detection method and device Download PDFInfo
- Publication number
- CN110688617A CN110688617A CN201910953212.4A CN201910953212A CN110688617A CN 110688617 A CN110688617 A CN 110688617A CN 201910953212 A CN201910953212 A CN 201910953212A CN 110688617 A CN110688617 A CN 110688617A
- Authority
- CN
- China
- Prior art keywords
- vibration
- fan
- working condition
- data set
- residual
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F03—MACHINES OR ENGINES FOR LIQUIDS; WIND, SPRING, OR WEIGHT MOTORS; PRODUCING MECHANICAL POWER OR A REACTIVE PROPULSIVE THRUST, NOT OTHERWISE PROVIDED FOR
- F03D—WIND MOTORS
- F03D17/00—Monitoring or testing of wind motors, e.g. diagnostics
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
Abstract
The invention discloses a method and a device for detecting abnormal vibration of a fan, wherein the method comprises the following steps: collecting fan operation data in a certain time period; extracting working condition characteristics and vibration characteristics from the fan operation data to obtain a test data set; predicting the vibration characteristics in the test data set by using a pre-established working condition-vibration model to obtain a predicted value corresponding to the vibration characteristics; calculating a vibration characteristic residual error according to the vibration characteristic and a predicted value thereof in the test data set to obtain a residual error matrix; calculating the distance of the vibration characteristic residual of each working condition point in the test data set according to the residual matrix; and if the distances of the vibration characteristic residuals of the working condition points of the continuously set number are all larger than a predetermined distance threshold, determining that the fan is abnormal in vibration in the time period. By the scheme of the invention, the fan vibration abnormity can be effectively detected under the condition that the high-frequency vibration signal of the fan component cannot be obtained.
Description
Technical Field
The invention relates to the field of data processing, in particular to a method and a device for detecting abnormal vibration of a fan.
Background
The existing wind turbine state monitoring system comprises fan vibration abnormity detection based on high-frequency vibration signals, for example, fault diagnosis is carried out in a Fourier transform and envelope spectrum analysis mode, and the detection needs a specific sensor to acquire the high-frequency vibration signals of corresponding parts. However, for the detection of the vibration abnormality of the components such as the tower drum And the nacelle of the wind turbine, usually, only low-frequency vibration signal Data from an SCADA (Supervisory Control And Data Acquisition) system can be collected, And if the existing high-frequency vibration signal abnormality detection method is used, effective fault characteristic frequency cannot be obtained from the low-frequency vibration signal Data, especially for a newly-built wind turbine generator or a generator with short effective operation time, And further, effective detection of the vibration abnormality of the wind turbine cannot be performed.
Disclosure of Invention
The embodiment of the invention provides a method and a device for detecting fan vibration abnormity, so that the fan vibration abnormity can be effectively detected even under the condition that a high-frequency vibration signal of a fan component cannot be obtained.
Therefore, the invention provides the following technical scheme:
a method for detecting abnormal vibration of a fan comprises the following steps:
collecting fan operation data in a certain time period;
extracting working condition characteristics and vibration characteristics from the fan operation data to obtain a test data set;
predicting the vibration characteristics in the test data set by using a pre-established working condition-vibration model to obtain a predicted value corresponding to the vibration characteristics;
calculating a vibration characteristic residual error according to the vibration characteristic and a predicted value thereof in the test data set to obtain a residual error matrix;
calculating the distance of the vibration characteristic residual of each working condition point in the test data set according to the residual matrix;
and if the distances of the vibration characteristic residuals of the working condition points of the continuously set number are all larger than a predetermined distance threshold, determining that the fan is abnormal in vibration in the time period.
Optionally, the method further comprises: establishing a working condition-vibration model according to historical data of the fan and determining a distance threshold, wherein the working condition-vibration model specifically comprises the following steps:
collecting historical data of the fan under different working conditions;
extracting working condition characteristics and vibration characteristics from the historical data of the fan to obtain a training data set;
and training by using the training data set to obtain a working condition-vibration model.
Optionally, the operating conditions include any one or more of: starting the fan, increasing the rotating speed/power, normally operating, reducing the load and operating, and stopping the fan.
Optionally, the extracting the operating condition characteristics and the vibration characteristics from the historical data of the fan includes:
filtering the historical data of the fan according to a set filtering rule;
segmenting the filtered historical data of the fan according to a time sequence to obtain one or more time sequences;
and for the time sequence, extracting the working condition characteristics and the vibration characteristics from each window in a sliding window mode.
Optionally, the condition-vibration model comprises a prediction model corresponding to each vibration characteristic; the method further comprises the following steps: determining a distance threshold value by using the working condition-vibration model and the training data set, specifically comprising:
determining the predicted value of each vibration feature in the training data set by using a prediction model corresponding to each vibration feature;
calculating a vibration characteristic residual error according to each vibration characteristic and a predicted value thereof in the training data set, and obtaining a residual error matrix when the distribution of the residual errors is in normal distribution;
calculating the distance of the vibration characteristic residual of each working condition point in the training data set according to the residual matrix;
and determining a distance threshold according to the distance of the vibration characteristic residual error of each working condition point.
A fan vibration abnormality detection apparatus, the apparatus comprising:
the data acquisition module is used for acquiring fan operation data in a certain time period;
the test data set generating module is used for extracting working condition characteristics and vibration characteristics from the fan operation data to obtain a test data set;
the prediction module is used for predicting the vibration characteristics in the test data set by utilizing a pre-established working condition-vibration model to obtain a predicted value corresponding to the vibration characteristics;
the residual error calculation module is used for calculating a vibration characteristic residual error according to the vibration characteristics and the predicted values thereof in the test data set to obtain a residual error matrix;
the distance calculation module is used for calculating the distance of the vibration characteristic residual of each working condition point in the test data set according to the residual matrix;
and the judging module is used for determining that the fan vibration is abnormal in the time period when the distances of the vibration characteristic residual errors of the working condition points of the continuously set number are all larger than a predetermined distance threshold value.
Optionally, the apparatus further comprises: the model establishing module is used for establishing a working condition-vibration model according to historical data of the fan; the model building module specifically comprises:
the data collection unit is used for collecting historical data of the fan under different working conditions;
the training data set generating unit is used for extracting working condition characteristics and vibration characteristics from the historical data of the fan to obtain a training data set;
and the training unit is used for training by using the training data set to obtain a working condition-vibration model.
Optionally, the operating conditions include any one or more of: starting the fan, increasing the rotating speed/power, normally operating, reducing the load and operating, and stopping the fan.
Optionally, the training data set generating unit includes:
the filtering subunit is used for filtering the historical data of the fan according to a set filtering rule;
the partitioning subunit is used for partitioning the filtered historical data of the fan according to a time sequence to obtain one or more time sequences;
and the characteristic extraction subunit is used for extracting the working condition characteristic and the vibration characteristic from each window by adopting a sliding window mode for the time sequence.
Optionally, the condition-vibration model comprises a prediction model corresponding to each vibration characteristic;
the device further comprises: a threshold determination module for determining a distance threshold using the working condition-vibration model and the training data set; the threshold determination module specifically includes:
the prediction unit is used for determining the prediction value of each vibration characteristic in the training data set by using a prediction model corresponding to each vibration characteristic;
the residual error calculation unit is used for calculating the residual errors of the vibration characteristics according to the vibration characteristics and the predicted values of the vibration characteristics in the training data set, and obtaining a residual error matrix when the distribution of the residual errors is in normal distribution;
the distance calculation unit is used for calculating the distance of the vibration characteristic residual of each working condition point in the training data set according to the residual matrix;
and the distance threshold value determining unit is used for determining the distance threshold value according to the distance of the vibration characteristic residual error of each working condition point.
An electronic device, comprising: one or more processors, memory;
the memory is configured to store computer-executable instructions and the processor is configured to execute the computer-executable instructions to implement the method described above.
A readable storage medium having stored thereon instructions which are executed to implement the foregoing method.
According to the method and the device for detecting the abnormal vibration of the fan, provided by the embodiment of the invention, a working condition-vibration model is established in advance according to low-frequency vibration signals of the fan under different working conditions, so that the defect that the working condition of a wind turbine generator set is complex in change is overcome; in addition, a uniform and quantitative measuring index is established for the difference of the vibration characteristics under different working conditions, and reference is provided for subsequent vibration abnormity detection. During online detection, collecting fan operation data within a certain time period, and extracting working condition characteristics and vibration characteristics from the data to obtain a test data set; then, predicting vibration characteristics in a test data set by using the working condition-vibration model to obtain a predicted value corresponding to the vibration characteristics; calculating a vibration characteristic residual error according to the vibration characteristic and a predicted value thereof in the test data set to obtain a residual error matrix; calculating the distance of each working condition point in the test data set according to the residual error matrix; and if the distances of the working condition points of the continuously set number are all larger than a predetermined distance threshold value, determining that the fan is abnormal in vibration in the time period. Thereby also can effectively detect fan vibration anomaly under the unable condition that obtains fan high frequency vibration signal, promoted the suitability of scheme.
Drawings
In order to more clearly illustrate the embodiments of the present application or technical solutions in the prior art, the drawings needed to be used in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments described in the present invention, and other drawings can be obtained by those skilled in the art according to the drawings.
FIG. 1 is a flow chart of a method for creating a condition-vibration model according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of partitioning a time sequence by using a sliding window method according to an embodiment of the present invention;
FIG. 3 is a flow chart of a method for detecting abnormal vibration of a fan according to an embodiment of the present invention;
FIG. 4 is a flow chart of determining a distance threshold in an embodiment of the present invention;
fig. 5 is a block diagram of a structure of a fan vibration abnormality detection apparatus according to an embodiment of the present invention;
FIG. 6 is a block diagram of a model building module according to an embodiment of the present invention;
FIG. 7 is a block diagram of a threshold determination module according to an embodiment of the present invention;
fig. 8 is another block diagram of the structure of the abnormal fan vibration detection apparatus according to the embodiment of the present invention.
Detailed Description
In order to make the technical field of the invention better understand the scheme of the embodiment of the invention, the embodiment of the invention is further described in detail with reference to the drawings and the implementation mode.
The embodiment of the invention provides a method and a device for detecting fan vibration abnormity, which are characterized in that a test data set is obtained by collecting fan operation data in a certain time period and extracting working condition characteristics and vibration characteristics from the data; then, predicting vibration characteristics in a test data set by using a pre-established working condition-vibration model to obtain a predicted value corresponding to the vibration characteristics; calculating a vibration characteristic residual error according to the vibration characteristic and a predicted value thereof in the test data set to obtain a residual error matrix; calculating the distance of each working condition point in the test data set according to the residual error matrix; and if the distances of the working condition points of the continuously set number are all larger than a predetermined distance threshold value, determining that the fan is abnormal in vibration in the time period.
It should be noted that the condition-vibration model includes a prediction model corresponding to each vibration characteristic, that is, a condition-vibration model needs to be established for each vibration characteristic.
First, a process of establishing a working condition-vibration model in the embodiment of the present invention will be described in detail.
As shown in fig. 1, it is a flowchart of establishing a working condition-vibration model in the embodiment of the present invention, and the method includes the following steps:
The working conditions comprise any one or more of the following: starting a fan, increasing the rotating speed/power, normally operating, reducing the load, stopping the fan and the like.
The historical data of the fan may specifically include, but is not limited to, any one or more of the following: time, power, rotating speed, wind speed, pitch angle, fan running state code, tower vibration signal and engine room vibration signal.
In practical application, in order to ensure the integrity of the working condition, the historical data of the fan may be data of at least 1 month of the fan operation. The frequency of use of the history data may be, for example, 1-5 s.
And 102, extracting working condition characteristics and vibration characteristics from the historical data of the fan to obtain a training data set.
In order to ensure the validity of data, the fan historical data can be firstly subjected to outlier detection, and operating condition points with obviously abnormal operating conditions are filtered out, wherein the operating condition points refer to the time points of data acquisition. That is, if a certain data is abnormal, all data of a point of time to which the data belongs are filtered out.
In the embodiment of the invention, the historical data of the fan can be filtered according to a set filtering rule. For example, the following filtering rules shown in table 1 may be set:
TABLE 1
| Variables of | Filtering rules |
| Wind speed | 0-30m/s |
| Power of | <Maximum design power of the fan |
| Wind speed-power | Wind power curve without deviating from unit design |
| Running state code of fan | Integer within specified range of wind field state code |
| Vibration letterNumber (C) | <Wind field vibration signal main control set threshold |
Then, segmenting the filtered historical data of the fan according to a time sequence to obtain one or more time sequences; for the time sequence, the time sequence is partitioned in a sliding window manner, such as shown in fig. 2. And for each time window data obtained by partitioning, respectively extracting the working condition characteristic and the vibration characteristic from each window.
Wherein the operating condition characteristics mainly include, but are not limited to, any one or more of the following: the average value, the peak-to-peak value, the maximum value, the standard deviation and the like of parameters such as wind speed, power, rotating speed, fan pitch angle and the like, and the mode value and the standard deviation of a fan running state code.
It should be noted that, because the fan operating status code is usually represented by a classification variable, in order to avoid the influence of the value on the modeling, one-hot encoding processing may be performed on the fan operating status code. One-Hot encoding is the representation of classification variables as binary vectors. This first requires mapping the classification values to integer values. Each integer value is then represented as a binary vector, which is a zero value, except for the index of the integer, which is marked as 1.
In the embodiment of the present invention, a manner of performing one-hot encoding on the fan operation status code may be as shown in table 2 below:
TABLE 2
Wherein the vibration characteristics mainly include, but are not limited to, any one or more of the following: the effective value, standard deviation, kurtosis value and entropy of the vibration signal.
Sequencing the extracted working condition features and the extracted vibration features according to a time sequence to obtain a training data set, wherein the training data can be expressed as: d: { DijTherein, 0<i<＝m，0<j<N; m is the number of working points, nIs the sum of the working condition characteristic and the vibration characteristic quantity.
And 103, training by using the training data set to obtain a working condition-vibration model.
In the embodiment of the present invention, it is necessary to separately establish a working condition-vibration model corresponding to each vibration characteristic, for example, a model corresponding to an effective value of the vibration signal, a model corresponding to a standard deviation of the vibration signal, a model corresponding to a kurtosis value of the vibration signal, and a model of an entropy of the vibration signal.
The working condition-vibration model can adopt a neural network model, an xgboost model and the like, and the training process of the model is similar to that of the existing corresponding model and is not described in detail herein.
Based on the working condition-vibration model, when the fan vibration is detected abnormally, the fan operation data in a certain time period can be collected, and the working condition characteristics and the vibration characteristics are extracted from the data to obtain a test data set. And predicting the vibration characteristics in the test data set by using the working condition-vibration model to obtain a predicted value corresponding to the vibration characteristics. And determining whether the fan vibration is abnormal or not according to the vibration characteristics in the test data set and the predicted value thereof.
As shown in fig. 3, it is a flowchart of a method for detecting abnormal vibration of a fan according to an embodiment of the present invention, and the method includes the following steps:
For example, data of 1 day of operation of the fan may be collected, and the sampling frequency may be, for example, 1-5 s. Of course, the time period and sampling frequency of the collected data may be set according to the actual application requirement, and the embodiment of the present invention is not limited thereto.
The fan operation data may specifically include, but is not limited to, any one or more of the following: time, power, rotating speed, wind speed, pitch angle, fan running state code, tower vibration signal and engine room vibration signal.
The extraction method of the working condition features and the vibration features can be referred to the description in the working condition-vibration model establishing process, and is not described herein again.
And 303, predicting the vibration characteristics in the test data set by using a pre-established working condition-vibration model to obtain a predicted value corresponding to the vibration characteristics.
As mentioned above, the condition-vibration model includes a prediction model for each vibration characteristic, and therefore, when predicting the vibration characteristics in the test data set, it is also necessary to predict each vibration characteristic using the prediction model corresponding to the vibration characteristic.
Specifically, the working condition characteristics related to the vibration characteristics in the test data set are input into the prediction model, and the prediction value of the vibration characteristics is obtained according to the output of the prediction model.
And 304, calculating a vibration characteristic residual error according to the vibration characteristics and the predicted values thereof in the test data set to obtain a residual error matrix.
The vibration characteristic residual error is the difference between the actual value of the acquired vibration characteristic and the predicted value obtained by using the prediction model. The residual matrix may be represented as: r: { RpqTherein, 0<p<＝m，0<q<N, m is the number of operating points, and n is the number of vibration characteristics.
And 305, calculating the distance of the vibration characteristic residual of each working condition point in the test data set according to the residual matrix.
It should be noted that, for different operating point residual error data, distances of corresponding vibration characteristic residual errors need to be calculated, and the distances may be mahalanobis distances and the like.
For example, when mahalanobis distance is used, the calculation formula is:
wherein, x is a residual vector of each working condition point in the residual matrix, mu is a mean vector of the residual of the working condition points, and Σ is a covariance matrix.
It should be noted that the mean vector and the covariance matrix in the above calculation formula can be obtained by using the aforementioned training data set in advance, and the specific calculation manner will be described in detail later.
And step 306, if the distances of the vibration characteristic residuals of the working condition points of the continuously set number are all larger than a predetermined distance threshold, determining that the fan vibration in the time period is abnormal.
Wherein the set number a may be determined as follows:
a is the duration of resonance/2/move of the fan;
wherein, move is the time step of each moving of the sliding window.
It should be noted that, in another embodiment of the method of the present invention, after the abnormal vibration of the fan is detected, an alarm may be given to remind a worker to repair the fan in time to remove the fault.
The distance threshold may be determined according to the aforementioned training data set and a pre-trained working condition-vibration model, and the specific flow is shown in fig. 4, and includes the following steps:
And 403, calculating the distance of the vibration characteristic residual of each working condition point in the training data set according to the residual matrix.
For example, for different operating point residual error data, the mahalanobis distances corresponding to the different operating point residual error data are respectively calculated, and the specific formula refers to the formula (1), wherein a mean vector and a covariance matrix in the formula are a mean vector and a covariance matrix of a residual error matrix of a training model.
And step 404, determining a distance threshold according to the distance of the vibration characteristic residual error of each working condition point.
For example, the maximum value may be selected from the calculated distances of the vibration characteristic residuals of the operating points as the distance threshold.
The following examples further illustrate the process of detecting abnormal vibration of the fan by using the solution of the embodiment of the present invention.
Taking the detection of the abnormal vibration of a tower drum of a certain wind field fan as an example, when a working condition-vibration model is established, historical operating data of the wind field fan for 1 month is collected, wherein the historical operating data comprises parameters such as time, power, rotating speed, wind speed, pitch angle, fan operating state codes, tower drum vibration signals and cabin vibration signals.
The working condition-vibration model training and distance threshold determining process is as follows:
1) and filtering the collected historical operating data of the fan, deleting part of outliers, and performing time series segmentation on the residual data.
2) And performing sliding window processing and relevant feature extraction on each time sequence, setting the size of a sliding window to be 5min, and setting the sliding moving step length to be 30s each time to obtain a training data set.
3) And respectively establishing four neural network models between the working condition characteristics and the single vibration characteristics, namely a prediction model corresponding to each vibration characteristic. The overall architecture of each neural network is as follows, and the network structure comprises four layers: input layer-first hidden layer-second hidden layer-output layer (the number of neurons is 18-10-5-1), the activation function of the hidden layer is tanh activation function, and the output layer has no activation function.
4) And calculating a vibration characteristic residual error according to the original value of the vibration characteristic in the training data set and the prediction results of the four prediction models, forming a residual error matrix and carrying out normality test in a QQ plot mode.
5) And calculating the Mahalanobis distance of the vibration characteristic residual errors of different working condition points, taking the maximum value of the Mahalanobis distance as a Mahalanobis distance threshold value, and storing a residual error mean vector and a covariance matrix.
The process of online detecting the fan vibration abnormity by utilizing the pre-established working condition-vibration model is as follows:
1) and (3) acquiring data of the wind field in actual operation for a certain day, and repeatedly executing the steps 2) to 3) in the model training process to obtain a test data set.
2) And predicting the vibration characteristics in the test data set by adopting a neural network model obtained by pre-training, and calculating the residual error of each vibration characteristic according to the prediction result to form a residual error matrix.
3) And calculating the Mahalanobis distance of each working condition point in the test data set through the Mahalanobis distance, wherein the mean vector in the Mahalanobis distance formula adopts the mean vector and the covariance matrix stored in the model training process.
4) Judging whether the Mahalanobis distance of continuous 5 working condition points exceeds the threshold value of the Mahalanobis distance; and if so, determining that the fan vibrates abnormally in the test time period.
According to the fan vibration abnormity detection method provided by the embodiment of the invention, a working condition-vibration model is established in advance according to low-frequency vibration signals of the fan under different working conditions, so that the defect that the working condition change of the wind turbine generator is complex is overcome; in addition, a uniform and quantitative measuring index is established for the difference of the vibration characteristics under different working conditions, and reference is provided for subsequent vibration abnormity detection. During online detection, collecting fan operation data within a certain time period, and extracting working condition characteristics and vibration characteristics from the data to obtain a test data set; then, predicting vibration characteristics in a test data set by using the working condition-vibration model to obtain a predicted value corresponding to the vibration characteristics; calculating a vibration characteristic residual error according to the vibration characteristic and a predicted value thereof in the test data set to obtain a residual error matrix; calculating the distance of each working condition point in the test data set according to the residual error matrix; and if the distances of the working condition points of the continuously set number are all larger than a predetermined distance threshold value, determining that the fan is abnormal in vibration in the time period. Thereby also can effectively detect fan vibration anomaly under the unable condition that obtains fan high frequency vibration signal, promoted the suitability of scheme.
Correspondingly, an embodiment of the present invention further provides a device for detecting abnormal vibration of a fan, as shown in fig. 5, the device includes the following modules:
the data acquisition module 501 is used for acquiring fan operation data within a certain time period;
a test data set generating module 502, configured to extract operating condition characteristics and vibration characteristics from the fan operating data to obtain a test data set;
the prediction module 503 is configured to predict the vibration characteristics in the test data set by using a pre-established working condition-vibration model 500, so as to obtain a prediction value corresponding to the vibration characteristics;
a residual calculation module 504, configured to calculate a vibration characteristic residual according to the vibration characteristic in the test data set and a predicted value thereof, so as to obtain a residual matrix;
a distance calculating module 505, configured to calculate, according to the residual error matrix, distances of vibration characteristic residual errors of the working condition points in the test data set;
and the judging module 506 is configured to determine that the fan vibration is abnormal within the time period when the distances of the vibration characteristic residuals of the working condition points of the continuously set number are all greater than a predetermined distance threshold.
It should be noted that the fan operation data may specifically include, but is not limited to, any one or more of the following: time, power, rotating speed, wind speed, pitch angle, fan running state code, tower vibration signal and engine room vibration signal. In practical applications, the data acquisition module 501 may acquire data of 1 day of operation of the wind turbine, for example, and the sampling frequency is 1-5 s. Of course, the time period and sampling frequency of the collected data may be set according to the actual application requirement, and the embodiment of the present invention is not limited thereto.
Specifically, the test data set generating module 502 may first perform outlier detection on the fan operating data, filter out operating condition points where operating conditions are obviously abnormal, and then segment the filtered fan operating data according to a time sequence to obtain one or more time sequences; and partitioning the time sequence in a sliding window mode. And for each time window data obtained by partitioning, respectively extracting the working condition characteristic and the vibration characteristic from each window.
Wherein the operating condition characteristics mainly include, but are not limited to, any one or more of the following: the mean value, peak-to-peak value, maximum value, standard deviation and the like of parameters such as wind speed, power, rotating speed, fan pitch angle and the like, and the mode value and standard deviation of a fan running state code; the vibration characteristics mainly include, but are not limited to, any one or more of the following: the effective value, standard deviation, kurtosis value and entropy of the vibration signal.
The distance of the vibration characteristic residual error of each working condition point may be a mahalanobis distance, and the specific calculation method has been described in detail above and is not described herein again.
The distance threshold may be determined based on the aforementioned training data set and a pre-trained condition-vibration model.
The working condition-vibration model may be constructed by a corresponding model building module, and the model building module may be a part of the apparatus of the present invention, or may be independent of the apparatus, and the embodiment of the present invention is not limited thereto.
As shown in fig. 6, it is a structural block diagram of a model building module in the embodiment of the present invention, and includes the following units:
the data collection unit 601 is used for collecting historical data of the fans under different working conditions;
a training data set generating unit 602, configured to extract operating condition features and vibration features from the historical data of the wind turbine to obtain a training data set;
and a training unit 603, configured to train to obtain a working condition-vibration model by using the training data set.
Wherein the working conditions comprise any one or more of the following: starting a fan, increasing the rotating speed/power, normally operating, reducing the load, stopping the fan and the like. The historical data of the fan may specifically include, but is not limited to, any one or more of the following: time, power, rotating speed, wind speed, pitch angle, fan running state code, tower vibration signal and engine room vibration signal.
In practical applications, in order to ensure the integrity of the working conditions, the historical data of the wind turbine collected by the data collection unit 601 may be data of at least 1 month of the operation of the wind turbine. The sampling frequency of the historical data may be, for example, 1-5s, and of course, other sampling frequencies may also be used, which is not limited in this embodiment of the present invention.
Specifically, the training data set generating unit 602 may first perform outlier detection on the historical data of the wind turbine, filter out operating condition points where operating conditions are obviously abnormal, and then segment the filtered historical data of the wind turbine according to a time sequence to obtain one or more time sequences; and partitioning the time sequence in a sliding window mode. And for each time window data obtained by partitioning, respectively extracting the working condition characteristic and the vibration characteristic from each window. Accordingly, one specific structure of the training data set generation unit 602 may include the following sub-units:
the filtering subunit is used for filtering the historical data of the fan according to a set filtering rule;
the partitioning subunit is used for partitioning the filtered historical data of the fan according to a time sequence to obtain one or more time sequences;
and the characteristic extraction subunit is used for extracting the working condition characteristic and the vibration characteristic from each window by adopting a sliding window mode for the time sequence.
As mentioned above, the distance threshold may be determined by the corresponding threshold determination module based on the aforementioned training data set and the pre-trained condition-vibration model. Similarly, the threshold determination module may be a part of the apparatus of the present invention, or may be independent of the apparatus, and the embodiment of the present invention is not limited thereto.
As shown in fig. 7, it is a structural block diagram of a threshold determining module in the embodiment of the present invention, and includes the following units:
a prediction unit 701, configured to determine a prediction value of each vibration feature in the training data set by using a prediction model corresponding to each vibration feature;
a residual calculation unit 702, configured to calculate a vibration characteristic residual according to each vibration characteristic and a predicted value thereof in the training data set, and obtain a residual matrix when the residual distribution is in a normal distribution;
a distance calculating unit 703, configured to calculate, according to the residual error matrix, a distance between residual errors of vibration characteristics of each operating point in the training data set;
and a distance threshold determining unit 704, configured to determine a distance threshold according to the distance of the vibration characteristic residual of each operating point.
According to the fan vibration abnormity detection device provided by the embodiment of the invention, a working condition-vibration model is established in advance according to low-frequency vibration signals of the fan under different working conditions, so that the defect that the working condition change of the wind turbine generator is complex is overcome; in addition, a uniform and quantitative measuring index is established for the difference of the vibration characteristics under different working conditions, and reference is provided for subsequent vibration abnormity detection. During online detection, collecting fan operation data within a certain time period, and extracting working condition characteristics and vibration characteristics from the data to obtain a test data set; then, predicting vibration characteristics in a test data set by using the working condition-vibration model to obtain a predicted value corresponding to the vibration characteristics; calculating a vibration characteristic residual error according to the vibration characteristic and a predicted value thereof in the test data set to obtain a residual error matrix; calculating the distance of each working condition point in the test data set according to the residual error matrix; and if the distances of the working condition points of the continuously set number are all larger than a predetermined distance threshold value, determining that the fan is abnormal in vibration in the time period. Thereby also can effectively detect fan vibration anomaly under the unable condition that obtains fan high frequency vibration signal, promoted the suitability of scheme.
Fig. 8 is a block diagram showing another structure of the abnormal fan vibration detection apparatus according to the embodiment of the present invention.
Unlike the embodiment shown in fig. 7, in this embodiment, the apparatus further includes: and the alarm module 801 is configured to alarm after the judgment module 506 determines that the fan is abnormal in vibration, so as to remind a worker to overhaul the fan in time and remove a fault.
It should be noted that, for the above embodiments of the device of the present invention, since the functional implementation of each module and unit is similar to that in the corresponding method, the description of the embodiments of the dialog generating device is relatively simple, and relevant points can be referred to the description of the corresponding parts of the embodiments of the method.
It should be noted that the terms "first," "second," and the like in the description and claims of the present invention and in the drawings described above are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used is interchangeable under appropriate circumstances such that the embodiments of the invention described herein are capable of operation in sequences other than those illustrated or described herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed, but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.
The embodiments in the present specification are described in a progressive manner, and the same and similar parts among the embodiments are referred to each other, and each embodiment focuses on the differences from the other embodiments. Furthermore, the above-described system embodiments are merely illustrative, wherein modules and units illustrated as separate components may or may not be physically separate, i.e., may be located on one network element, or may be distributed over a plurality of network elements. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the present embodiment. One of ordinary skill in the art can understand and implement it without inventive effort.
Those skilled in the art will appreciate that all or part of the steps in the above method embodiments may be implemented by a program to instruct relevant hardware to perform the steps, and the program may be stored in a computer-readable storage medium, referred to herein as a storage medium, such as: ROM/RAM, magnetic disk, optical disk, etc.
Correspondingly, the embodiment of the invention also provides a device for the fan vibration abnormity detection method, and the device is an electronic device, such as a mobile terminal, a computer, a tablet device, a medical device, a fitness device, a personal digital assistant and the like. The electronic device may include one or more processors, memory; wherein the memory is used for storing computer executable instructions and the processor is used for executing the computer executable instructions to realize the method of the previous embodiments.
The present invention has been described in detail with reference to the embodiments, and the description of the embodiments is provided to facilitate the understanding of the method and apparatus of the present invention, and is intended to be a part of the embodiments of the present invention rather than the whole embodiments. All other embodiments obtained by a person of ordinary skill in the art based on the embodiments of the present invention without any creative effort shall fall within the protection scope of the present invention, and the content of the present description shall not be construed as limiting the present invention. Therefore, any modification, equivalent replacement, improvement and the like made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (10)
1. A method for detecting abnormal vibration of a fan is characterized by comprising the following steps:
collecting fan operation data in a certain time period;
extracting working condition characteristics and vibration characteristics from the fan operation data to obtain a test data set;
predicting the vibration characteristics in the test data set by using a pre-established working condition-vibration model to obtain a predicted value corresponding to the vibration characteristics;
calculating a vibration characteristic residual error according to the vibration characteristic and a predicted value thereof in the test data set to obtain a residual error matrix;
calculating the distance of the vibration characteristic residual of each working condition point in the test data set according to the residual matrix;
and if the distances of the vibration characteristic residuals of the working condition points of the continuously set number are all larger than a predetermined distance threshold, determining that the fan is abnormal in vibration in the time period.
2. The method of claim 1, further comprising: establishing a working condition-vibration model according to historical data of the fan and determining a distance threshold, wherein the working condition-vibration model specifically comprises the following steps:
collecting historical data of the fan under different working conditions;
extracting working condition characteristics and vibration characteristics from the historical data of the fan to obtain a training data set;
and training by using the training data set to obtain a working condition-vibration model.
3. The method of claim 2, wherein the operating conditions include any one or more of: starting the fan, increasing the rotating speed/power, normally operating, reducing the load and operating, and stopping the fan.
4. The method of claim 2, wherein extracting operating condition features and vibration features from the historical wind turbine data comprises:
filtering the historical data of the fan according to a set filtering rule;
segmenting the filtered historical data of the fan according to a time sequence to obtain one or more time sequences;
and for the time sequence, extracting the working condition characteristics and the vibration characteristics from each window in a sliding window mode.
5. The method of claim 2, wherein the condition-vibration model comprises a predictive model for each vibration characteristic; the method further comprises the following steps: determining a distance threshold value by using the working condition-vibration model and the training data set, specifically comprising:
determining the predicted value of each vibration feature in the training data set by using a prediction model corresponding to each vibration feature;
calculating a vibration characteristic residual error according to each vibration characteristic and a predicted value thereof in the training data set, and obtaining a residual error matrix when the distribution of the residual errors is in normal distribution;
calculating the distance of the vibration characteristic residual of each working condition point in the training data set according to the residual matrix;
and determining a distance threshold according to the distance of the vibration characteristic residual error of each working condition point.
6. A fan vibration anomaly detection device, said device comprising:
the data acquisition module is used for acquiring fan operation data in a certain time period;
the test data set generating module is used for extracting working condition characteristics and vibration characteristics from the fan operation data to obtain a test data set;
the prediction module is used for predicting the vibration characteristics in the test data set by utilizing a pre-established working condition-vibration model to obtain a predicted value corresponding to the vibration characteristics;
the residual error calculation module is used for calculating a vibration characteristic residual error according to the vibration characteristics and the predicted values thereof in the test data set to obtain a residual error matrix;
the distance calculation module is used for calculating the distance of the vibration characteristic residual of each working condition point in the test data set according to the residual matrix;
and the judging module is used for determining that the fan vibration is abnormal in the time period when the distances of the vibration characteristic residual errors of the working condition points of the continuously set number are all larger than a predetermined distance threshold value.
7. The apparatus of claim 6, further comprising: the model establishing module is used for establishing a working condition-vibration model according to historical data of the fan; the model building module specifically comprises:
the data collection unit is used for collecting historical data of the fan under different working conditions;
the training data set generating unit is used for extracting working condition characteristics and vibration characteristics from the historical data of the fan to obtain a training data set;
and the training unit is used for training by using the training data set to obtain a working condition-vibration model.
8. The apparatus of claim 7, wherein the operating conditions include any one or more of: starting the fan, increasing the rotating speed/power, normally operating, reducing the load and operating, and stopping the fan.
9. The apparatus of claim 7, wherein the training data set generation unit comprises:
the filtering subunit is used for filtering the historical data of the fan according to a set filtering rule;
the partitioning subunit is used for partitioning the filtered historical data of the fan according to a time sequence to obtain one or more time sequences;
and the characteristic extraction subunit is used for extracting the working condition characteristic and the vibration characteristic from each window by adopting a sliding window mode for the time sequence.
10. The apparatus of claim 7, wherein the condition-vibration model comprises a predictive model for each vibration characteristic;
the device further comprises: a threshold determination module for determining a distance threshold using the working condition-vibration model and the training data set; the threshold determination module specifically includes:
the prediction unit is used for determining the prediction value of each vibration characteristic in the training data set by using a prediction model corresponding to each vibration characteristic;
the residual error calculation unit is used for calculating the residual errors of the vibration characteristics according to the vibration characteristics and the predicted values of the vibration characteristics in the training data set, and obtaining a residual error matrix when the distribution of the residual errors is in normal distribution;
the distance calculation unit is used for calculating the distance of the vibration characteristic residual of each working condition point in the training data set according to the residual matrix;
and the distance threshold value determining unit is used for determining the distance threshold value according to the distance of the vibration characteristic residual error of each working condition point.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910953212.4A CN110688617B (en) | 2019-10-09 | 2019-10-09 | Fan vibration abnormity detection method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910953212.4A CN110688617B (en) | 2019-10-09 | 2019-10-09 | Fan vibration abnormity detection method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110688617A true CN110688617A (en) | 2020-01-14 |
| CN110688617B CN110688617B (en) | 2020-07-03 |
Family
ID=69111600
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910953212.4A Active CN110688617B (en) | 2019-10-09 | 2019-10-09 | Fan vibration abnormity detection method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110688617B (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111047121A (en) * | 2020-03-17 | 2020-04-21 | 浙江上风高科专风实业有限公司 | Fan fault prediction method and system |
| CN111275416A (en) * | 2020-01-15 | 2020-06-12 | 中国人民解放军国防科技大学 | Digital currency abnormal transaction detection method and device, electronic equipment and medium |
| CN112360702A (en) * | 2020-11-16 | 2021-02-12 | 华能昌邑风力发电有限公司 | Centralized monitoring method and device for vibration data |
| CN113281003A (en) * | 2021-06-18 | 2021-08-20 | 浙江华东测绘与工程安全技术有限公司 | Fixed marine structure real-time damage monitoring method and load level monitoring method |
| CN113719425A (en) * | 2021-05-14 | 2021-11-30 | 北京智慧空间科技有限责任公司 | Fan blade fault prediction method and system and storage medium |
| CN113982850A (en) * | 2021-09-10 | 2022-01-28 | 大唐国信滨海海上风力发电有限公司 | Fan comprehensive health analysis method and system fusing high-frequency and low-frequency signals |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105719002A (en) * | 2016-01-18 | 2016-06-29 | 重庆大学 | Wind turbine generator state parameter abnormity identification method based on combination prediction |
| US20190226943A1 (en) * | 2018-01-23 | 2019-07-25 | Mitsubishi Heavy Industries, Ltd. | Facility condition monitoring device and facility condition monitoring method |
| CN110259646A (en) * | 2019-05-06 | 2019-09-20 | 明阳智慧能源集团股份公司 | A kind of wind power generating set unit status method for early warning based on historical data |
-
2019
- 2019-10-09 CN CN201910953212.4A patent/CN110688617B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105719002A (en) * | 2016-01-18 | 2016-06-29 | 重庆大学 | Wind turbine generator state parameter abnormity identification method based on combination prediction |
| US20190226943A1 (en) * | 2018-01-23 | 2019-07-25 | Mitsubishi Heavy Industries, Ltd. | Facility condition monitoring device and facility condition monitoring method |
| CN110259646A (en) * | 2019-05-06 | 2019-09-20 | 明阳智慧能源集团股份公司 | A kind of wind power generating set unit status method for early warning based on historical data |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111275416A (en) * | 2020-01-15 | 2020-06-12 | 中国人民解放军国防科技大学 | Digital currency abnormal transaction detection method and device, electronic equipment and medium |
| CN111275416B (en) * | 2020-01-15 | 2024-02-27 | 中国人民解放军国防科技大学 | Digital currency abnormal transaction detection method, device, electronic equipment and medium |
| CN111047121A (en) * | 2020-03-17 | 2020-04-21 | 浙江上风高科专风实业有限公司 | Fan fault prediction method and system |
| CN112360702A (en) * | 2020-11-16 | 2021-02-12 | 华能昌邑风力发电有限公司 | Centralized monitoring method and device for vibration data |
| CN112360702B (en) * | 2020-11-16 | 2022-11-25 | 华能昌邑风力发电有限公司 | Centralized monitoring method and device for vibration data |
| CN113719425A (en) * | 2021-05-14 | 2021-11-30 | 北京智慧空间科技有限责任公司 | Fan blade fault prediction method and system and storage medium |
| CN113719425B (en) * | 2021-05-14 | 2022-11-22 | 北京智慧空间科技有限责任公司 | Fan blade fault prediction method and system and storage medium |
| CN113281003A (en) * | 2021-06-18 | 2021-08-20 | 浙江华东测绘与工程安全技术有限公司 | Fixed marine structure real-time damage monitoring method and load level monitoring method |
| CN113982850A (en) * | 2021-09-10 | 2022-01-28 | 大唐国信滨海海上风力发电有限公司 | Fan comprehensive health analysis method and system fusing high-frequency and low-frequency signals |
| CN113982850B (en) * | 2021-09-10 | 2023-11-10 | 大唐国信滨海海上风力发电有限公司 | Fan comprehensive health analysis method and system integrating high-low frequency signals |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110688617B (en) | 2020-07-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110688617B (en) | Fan vibration abnormity detection method and device | |
| CN109726524B (en) | CNN and LSTM-based rolling bearing residual service life prediction method | |
| Li et al. | Deep learning-based remaining useful life estimation of bearings using multi-scale feature extraction | |
| CN106104496B (en) | The abnormality detection not being subjected to supervision for arbitrary sequence | |
| WO2019080367A1 (en) | Method for evaluating health status of mechanical device | |
| CN103115789B (en) | Second generation small-wave support vector machine assessment method for damage and remaining life of metal structure | |
| CN111400961B (en) | Wind generating set blade fault judging method and device | |
| CN112766342A (en) | Abnormity detection method for electrical equipment | |
| CN112629585A (en) | Equipment on-line monitoring method and device based on multi-dimensional parameter estimation | |
| CN110704964A (en) | Steam turbine operation state diagnosis method and device, electronic equipment and storage medium | |
| CN110909826A (en) | Diagnosis monitoring method and device for energy equipment and electronic equipment | |
| CN111191838A (en) | Industrial equipment state control method and device integrating artificial intelligence algorithm | |
| CN116306798A (en) | Ultra-short time wind speed prediction method and system | |
| CN113110961B (en) | Equipment abnormality detection method and device, computer equipment and readable storage medium | |
| CN114048879A (en) | Fault detection method, device, system, computer equipment and readable storage medium | |
| Boukra | Identifying new prognostic features for remaining useful life prediction using particle filtering and neuro-fuzzy system predictor | |
| KR102108975B1 (en) | Apparatus and method for condition based maintenance support of naval ship equipment | |
| CN116625683A (en) | Wind turbine generator system bearing fault identification method, system and device and electronic equipment | |
| CN115859791A (en) | Method and device for estimating residual life of aero-engine | |
| CN114881321A (en) | Mechanical component failure prediction method, device, electronic device and storage medium | |
| Kumar et al. | Gas turbine engine operational data analysis for anomaly detection: Statistical vs. neural network approach | |
| Hoffmann et al. | Embedding Anomaly Detection Autoencoders for Wind Turbines | |
| CN111985380B (en) | Bearing degradation process state monitoring method, system, equipment and storage medium | |
| KR102212022B1 (en) | Method of automatically determining condition of hydro turbine in hydroelectric power plant and system for the same | |
| CN117409816B (en) | Equipment fault detection method and system based on sound signals |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |