CN110674891B - Data quality abnormity detection method for monitoring system - Google Patents
Data quality abnormity detection method for monitoring system Download PDFInfo
- Publication number
- CN110674891B CN110674891B CN201910984189.5A CN201910984189A CN110674891B CN 110674891 B CN110674891 B CN 110674891B CN 201910984189 A CN201910984189 A CN 201910984189A CN 110674891 B CN110674891 B CN 110674891B
- Authority
- CN
- China
- Prior art keywords
- data
- vibration data
- vibration
- abnormal
- value
- 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.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/2433—Single-class perspective, e.g. one-against-all classification; Novelty detection; Outlier detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/21—Design, administration or maintenance of databases
- G06F16/215—Improving data quality; Data cleansing, e.g. de-duplication, removing invalid entries or correcting typographical errors
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Quality & Reliability (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
- Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)
Abstract
The application relates to a data quality abnormity detection method of a monitoring system, which comprises vibration data drift abnormity detection, vibration data interruption abnormity detection, vibration data singular value abnormity detection, vibration data multipoint repetition abnormity detection, vibration data positive and negative data point difference value proportion detection, vibration data beyond sensor range abnormity detection, rotating speed pulse data abnormity detection, vibration data and rotating speed pulse data length abnormity detection and vibration data and rotating speed pulse data missing value abnormity detection, and provides corresponding detection methods for the detection models, so that the data abnormity detection method can efficiently and accurately detect the data abnormity value of the monitoring system.
Description
Technical Field
The application relates to a method for detecting data quality abnormity of a monitoring system, which is suitable for the technical field of industrial equipment monitoring.
Background
Industrial equipment is equipment for industrial production, and the health condition of the equipment directly influences whether the industrial production can be smoothly carried out. In practice, as the service time of the equipment increases, the failure rate of the equipment gradually increases, and normal production is affected. Some major equipment with high economic value, such as wind turbine generators, gas turbine generators, large compressors and the like, can bring great economic loss when in failure, and even can cause safety accidents to cause personnel to finish the work. Therefore, the operation and maintenance work of the industrial equipment is a key ring in industrial production, and the economic benefit and the production safety of the industrial production are directly influenced.
The traditional operation and maintenance of industrial equipment usually adopts a regular inspection mode, and the maintenance mode has the defects of over maintenance and under maintenance. In order to solve the above problems, many process equipment manufacturers add a "state monitoring system", referred to as "CMS" for short, to industrial equipment. CMS mainly installs vibration sensor and revolution speed transducer additional on parts such as industrial equipment's bearing, gear box, generator and carries out real-time supervision to industrial equipment part, through the whole health status to each core part health status of vibration data analysis recognizable equipment and equipment to rationally arrange the fortune dimension plan according to the health status of part, avoided traditional fortune dimension mode to maintain and lack the maintenance shortcoming. In practice, CMS data often causes abnormality of vibration data and rotation speed data due to sensor damage, sensor drop, network transmission abnormality, edge data acquisition hardware defect, and the like. The vibration data and the abnormal rotating speed data directly influence the vibration analysis result, so that the CMS reports faults in a wrong way or fails to report faults, and bad experience and trouble are brought to operation and maintenance personnel. Therefore, CMS data quality abnormity is detected, data quality abnormity prompt is given in time, misjudgment of field operation and maintenance personnel on the CMS diagnosis result due to data quality problems can be avoided, and the method has great significance on CMS actual application.
There are also some related studies on the aspect of data quality anomaly detection, such as performing stationarity, normality and periodicity tests on data to identify whether the data is abnormal or not. The method has better results in the detection of the data abnormality with stable periodicity and stable distribution, but is not suitable for the detection of the complex vibration data abnormality in the CMS. The CMS vibration data is non-stationary data, and the periodicity and the data distribution characteristics of the vibration data are different when different faults occur to equipment, so that the detection cannot be performed by using single data stationarity, periodicity and normality. In addition, a learner may calculate a deviation between a predicted value and an actual value of the vibration data using an AR prediction model to detect whether the data is abnormal. Due to the complexity and non-stationarity of CMS vibration data, the AR model is not suitable for prediction of CMS vibration data. Also, since the vibration data corresponding to different faults are very different, the CMS vibration data is difficult to predict in time series.
The data quality anomaly detection by using the differential method is also applied in practice, for example, the vibration data is detected by using a first-order differential method, and whether the data quality is abnormal or not is identified by identifying the mutation value in the differential sequence or comparing the distribution consistency of the differential sequence of different sensors and different time periods. The method is suitable for abnormal detection scenes with unobvious data impact characteristics, slow change or large data correlation among different sensors. However, in the industrial equipment state monitoring system, the correlation of vibration data among the vibration sensors is small, the vibration data changes rapidly, and the vibration data itself has an impact characteristic when a bearing or a gear or other parts are in failure, so that the method is not suitable for abnormal detection of the vibration data of the industrial equipment. The data quality anomaly detection based on the correlation principle is also applied in practice, for example, the data quality is considered to be abnormal when the data of different sensors are greatly different by comparing the data of different sensors; or whether the data quality is abnormal or not is identified by calculating the linear correlation coefficient of the sensor data and the actual load data. The data quality abnormity detection method is established on the principle of correlation, is not suitable for the scene of poor correlation among vibration data of industrial equipment such as motors, gear boxes, fans and the like, and can fail when a plurality of sensors have data abnormity at the same time. In addition, the method for detecting the data quality abnormity by comparing whether the two moments before and after the data have obvious difference is also an existing data quality abnormity detection method. The method is suitable for detecting abnormal data quality of slowly-varying data, such as temperature, humidity and the like, and is not suitable for vibration data with vibration amplitude changing rapidly at any moment. In addition, whether the data is abnormal or not is identified by judging information such as amplitude, frequency and the like of the data by a method of purely analyzing a frequency spectrum. The data anomaly detection method is based on the frequency component specificity in the data, and when a specific frequency component is detected to be anomalous, the data is considered to be anomalous. The frequency components of the vibration data are complex, and when the data is abnormal, it cannot be predicted in advance which frequency components are abnormal, so that the method of simply judging whether a certain frequency component is abnormal is not suitable for detecting the abnormal vibration data. In addition, the prior art can not well identify abnormal conditions such as vibration data interruption, data drift, data length missing and the like, and can not judge the abnormal rotating equipment rotating speed pulse data. Therefore, the existing data quality abnormity detection method has defects in the data quality detection of the industrial equipment state monitoring system, and can not meet the data quality abnormity detection requirement of the industrial equipment state monitoring system.
Disclosure of Invention
The invention aims to provide a data quality abnormity detection method of a monitoring system, which can realize the abnormity detection of vibration data and rotating speed pulse data of industrial equipment such as a motor, a gear box, a fan and the like, particularly realize the detection of various possible data abnormity, and provide a detection method with high calculation efficiency and high accuracy so as to realize the purpose of the application.
According to the data quality abnormity detection method of the monitoring system, at least one of vibration data drift abnormity detection and vibration data interruption abnormity detection is carried out, and a vibration data sequence X is obtained by arranging a vibration sensor1,x2,x3,…,xn]Wherein x isi(i-1 … n) is the data point in the vibration sequence, n is the data length; the abnormal drift detection of the vibration data adopts a sliding window to calculate the mean value of the vibration data in a segmentation way, and takes the maximum value M of the absolute value of the mean value of each section of the vibration dataaBy judging MaWhether the vibration data deviate from the 0 line is identified;
the step of detecting the abnormal drifting of the vibration data is as follows:
1) setting the length N of a sliding window and a sliding step length s, wherein s is more than 0 and less than or equal to N;
2) sliding the sliding window backwards and successively by step length s from the starting time of the vibration data, and respectively calculating the absolute value M of the mean value of the vibration data in the sliding window after each sliding as [ M [ ]1,m2,m3,…,mK]Wherein, in the step (A),
(i ═ 0 … K, K is the number of window slides);
3) calculating the maximum value M of Ma＝max(M)；
4) Comparison MaAnd a predetermined vibration data drift threshold th, when M isa>th, judging that the quality of the vibration data is abnormal;
the vibration data interruption anomaly detection adopts a sliding window to calculate the effective value of the vibration data in a segmented mode, and whether interruption anomaly exists in the vibration data is detected by judging whether the maximum effective value max _ r and the minimum effective value min _ r are greatly different; the step of detecting the abnormal interruption of the vibration data is as follows:
1) setting the length N of a sliding window and a sliding step length s, wherein s is more than 0 and less than or equal to N; the length of the data sliding window of each vibration measuring point meets the condition: n is more than or equal to T multiplied by fs, wherein T is the rotating period of the rotating shaft corresponding to the vibration measuring point, and fs is the sampling frequency;
2) sliding the sliding window backwards and successively by step length s from the starting time of the vibration data, and respectively calculating the effective value R of the vibration data in the sliding window after each sliding as [ R ═ R1,r2,r3,…,rK]Wherein, in the step (A),
(i ═ 0 … K, K is the number of window slides);
3) calculating min _ r ═ min (r), max _ r ═ max (r);
4) computing
Comparing u with a preset data interruption threshold th, when u is greater than th>And th, judging that the vibration data is abnormal.
Preferably, the method further comprises the step of abnormal detection of singular values of the vibration data, and the abnormal detection of singular values of the vibration data comprises the following steps:
1) calculate root mean square of X:
2) calculating X1＝abs(X)；
3) Setting a root-mean-square coefficient k;
4) statistics of X1The number num of data which is larger than k multiplied by sd is calculated, and the singular value ratio P is equal to num/n;
5) and setting a singular value anomaly detection threshold th, comparing the sizes of P and th, and judging that the quality of the vibration data is abnormal when P is greater than th.
Preferably, the method further comprises the step of detecting multipoint repeated abnormity of the vibration data, and the step of detecting multipoint repeated abnormity of the vibration data comprises the following steps:
1) calculating a unique numerical sequence Y ═ Y [ Y ] that is not repeated in the vibration data X1,y2,y3,…,yk]；
2) The number of times R _ num ═ m is calculated for each data point in Y that recurs in X1,m2,m3,…,mk]Wherein m isi(i-1 … k) calculating the maximum value max _ num-max (R _ num) of the repeated occurrences of each data point;
3) setting a repeated data point threshold th, comparing the size of max _ num with the size of th, and judging that the quality of the vibration data is abnormal when max _ num is greater than th.
Preferably, the method further comprises detecting the positive and negative data point number difference ratio of the vibration data, and the detecting step is as follows:
1) calculating vibration data points P1 greater than 0 and data points P2 less than 0 in the vibration data X;
2) calculating s ═ abs (P1-P2)/n;
3) setting a positive and negative data point number difference ratio threshold th;
4) and comparing the sizes of s and th, and judging that the quality of the vibration data is abnormal when s is larger than th.
Preferably, the method further comprises the step of detecting abnormal vibration data exceeding the range of the sensor, wherein the detection step is as follows:
1) calculating a peak value P _ P ═ max (X) -min (X) of each time the edge data acquisition equipment uploads the vibration data;
2) comparing the size of P _ P and 2A, wherein A is the measuring range of the sensor; and if P _ P is greater than 2A, judging that the vibration data exceeds the range of the sensor.
Preferably, the method further comprises the step of detecting the abnormal rotating speed pulse data, wherein the rotating speed pulse data is obtained by arranging a rotating speed sensor, and the detecting of the abnormal rotating speed pulse data comprises the following specific steps:
1) extracting a time sequence t ═ t corresponding to the rising edge position of the rotating speed pulse data1,t2,,…,tn]；
2) Assuming that P pulses are generated in each rotation under normal conditions, sampling is carried out on T every P-1 points, and a time sequence T ═ T [ T ] corresponding to the rising edge of the rotating speed pulse in the whole period can be obtained1,T2,,…,Tn]；
3) Calculating each whole periodCorresponding rotational speed spi＝60/Ti(i＝1,2,…,n)；
4) Comparing sp one by oneiAnd the maximum rotation speed max _ r and minimum rotation speed min _ r of the device, when spi>max _ r or spi<When min _ r is needed, judging that the rotating speed pulse data are abnormal; alternatively, the rotation speed fluctuation coefficient is calculated
If s>th, judging that the rotating speed pulse data is abnormal; wherein th is a preset threshold value of the rotation speed fluctuation coefficient.
Preferably, the method further comprises the step of detecting abnormal lengths of the vibration data and the rotating speed pulse data, and the step comprises the following steps:
1) calculating the lengths of vibration data and rotating speed pulse data of each time of uploading data by the edge end as vib _ n and imp _ n respectively;
2) judging whether the vib _ n is equal to the preset data length or not, and if not, judging that the vibration data quality is abnormal;
and judging whether the imp _ n is equal to the preset data length or not, and if not, judging that the quality of the rotating speed pulse data is abnormal.
Preferably, the method further comprises the step of detecting the missing value abnormality of the vibration data and the rotation speed pulse data, namely judging whether the value corresponding to each data point is a null value or not point by point aiming at the vibration data and the rotation speed pulse data sequence uploaded at each time of the edge end, and judging that the vibration data or the rotation speed pulse data are abnormal when one or more null values appear in the vibration data sequence and the rotation speed pulse data sequence.
The method and the device can realize abnormal detection of the drifting of the vibration data and abnormal detection of the interruption of the vibration data caused by poor connection of the sensor cable or temperature influence, and can also realize common data loss and abnormal detection of singular values of the vibration data. Meanwhile, the method can realize abnormal detection of the problem that the vibration data exceeds the range of the sensor due to damage of the sensor, and abnormal detection can be identified by the method due to loss of the vibration data or the rotating speed pulse data, multipoint repetition of the vibration data, and the number difference ratio of the positive data and the negative data of the vibration data caused by network reasons, loose connection of the cable of the sensor, edge data acquisition hardware and the like. The method designed by the invention can also be used for carrying out abnormity detection on the abnormal length of the vibration data and can be used for detecting the abnormal speed pulse data. The data anomaly detection method for the equipment state monitoring system can almost identify all conditions of data quality anomaly of the state monitoring system, has wide coverage range and strong practicability, and is particularly suitable for data quality anomaly detection of the equipment state monitoring system.
Another aspect of the present application provides a corresponding detection method for several detection models, such as detection of abnormal drifting of vibration data, detection of abnormal interruption of vibration data, detection of abnormal singular value of vibration data, detection of abnormal multipoint repetition of vibration data, detection of ratio of positive and negative data point difference of vibration data, detection of abnormal span of vibration data beyond a sensor, detection of abnormal rotating speed pulse data, detection of abnormal length of vibration data and rotating speed pulse data, and detection of abnormal missing value of vibration data and rotating speed pulse data, which can efficiently and accurately detect abnormal data values.
Drawings
FIG. 1 is a schematic illustration of vibration data interruption as described herein.
Fig. 2 is a diagram showing a waveform of data of the multi-point repetitive vibration in the embodiment.
Fig. 3 shows a simulated waveform diagram of vibration data drift in the embodiment.
Fig. 4 shows a simulated waveform diagram of vibration data interruption in an embodiment.
Fig. 5 is a waveform diagram showing simulation in which singular values exist in vibration data in the embodiment.
Fig. 6 is a waveform diagram showing a simulation of abnormal revolution speed pulse data in the embodiment.
FIG. 7 is a schematic diagram showing the rotational speed information calculated from the rotational speed pulse data of FIG. 6.
Detailed Description
To make the objects, technical solutions and advantages of the present application more apparent, embodiments of the present application will be described in detail below with reference to the accompanying drawings. It should be noted that the embodiments and features of the embodiments in the present application may be arbitrarily combined with each other without conflict.
The vibration sensor and the rotating speed sensor are additionally arranged on industrial equipment such as a motor, a gear box and a fan to monitor each part in real time, the health condition of each part can be identified by analyzing vibration data, and an operation and maintenance plan is reasonably arranged according to the health state of a large part.
The data quality abnormity detection method of the monitoring system comprises at least one of vibration data drifting abnormity detection and vibration data interruption abnormity detection. Preferably, the method can also comprise vibration data multipoint repetition abnormity detection and/or vibration data singular value abnormity detection. More preferably, the method further comprises the steps of detecting missing values of the vibration data and the rotating speed pulse data, detecting abnormal ratio of positive and negative data points of the vibration data, detecting abnormal length of the vibration data and the rotating speed pulse data, detecting abnormal range of the vibration data beyond a sensor, detecting abnormal rotating speed pulse data and the like. The data anomaly detection basically covers the data anomaly with the occurrence probability of more than 95% in practice, and can meet the requirement on the anomaly detection of the state data of the industrial equipment.
The application also relates to a specific judgment method for the data abnormity, the judgment method can improve the accuracy, the speed and the effectiveness of judgment, and the algorithm is simple and easy to implement. Specific methods of the above-described various data abnormality detection will be described separately below.
Singular value anomaly detection for vibration data
Singular values with outstanding amplitude values often appear in the vibration data, and excessive singular values can influence the analysis result of the vibration data. Therefore, whether the vibration data is abnormal is identified by determining whether there are excessive singular values in the vibration data. Suppose the vibration data sequence X ═ X1,x2,x3,…,xn]Wherein x isi(i-1 … n) is the data point in the vibration sequence and n is the data length. The singular value anomaly detection method of the vibration data comprises the following steps:
1) calculate root mean square of X:
2) calculating X1Abs (x), where abs (.) is the symbol of the absolute value;
3) setting a root-mean-square coefficient k, wherein the value of k can be between 3 and 8, and the smaller the value is, the stricter the abnormal detection is;
4) statistics of X1The number num of data which is larger than k multiplied by sd is calculated, and the singular value ratio P is equal to num/n;
5) setting a singular value anomaly detection threshold value th, wherein the value of th can be between 0.00005 and 0.001 in practice, and the smaller the value of th is, the stricter the anomaly detection is;
6) comparing the sizes of P and th, and judging that the quality of the vibration data is abnormal when P is larger than th.
Vibration data drift anomaly detection
Vibration data drift is usually caused by reasons such as loose connecting wires of the vibration sensors or extreme environmental temperature, the vibration data of the whole duration or the center of a certain section of vibration data can deviate from a zero line, and the vibration data are abnormal. The invention adopts a sliding window to calculate the average value of the vibration data in a sectional way and takes the maximum value M of the absolute value of the average value of each section of the vibration dataaBy judging MaWhether or not the 0 line is deviated to identify whether or not the vibration data has drifted. Suppose the vibration data sequence X ═ X1,x2,x3,…,xn]Wherein x isi(i-1 … n) is the data point in the vibration sequence and n is the data length. The vibration data drift abnormity detection steps are as follows:
1) setting the length N of a sliding window and a sliding step length s, wherein s is more than 0 and less than or equal to N;
2) sliding the sliding window backwards and successively by step length s from the starting time of the vibration data, and respectively calculating the absolute value M of the mean value of the vibration data in the sliding window after each sliding as [ M [ ]1,m2,m3,…,mK]Wherein, in the step (A),
(i is 0 … K, K is the number of window slides, abs (.) is the symbol of the absolute value);
3) calculating the maximum value M of Ma＝max(M)；
4) Comparison MaAnd magnitude of vibration data drift threshold th, when Ma>And th, judging that the quality of the vibration data is abnormal. Aiming at the vibration data of the industrial equipment, th value can be between 0.1m/s2Between-1 m/s, the smaller the th value, the stricter the abnormal detection is.
Vibration data interrupt anomaly detection
When the vibration sensor is not firmly connected with the sensor cable or the sensor cable is not firmly connected with the edge data acquisition equipment, the cable shake caused by vibration may cause the sensor or the edge data acquisition equipment to be disconnected with the cable for a short time, and finally the vibration data is interrupted, wherein the typical vibration data interruption is characterized in that the amplitude of the vibration data is suddenly and greatly reduced within a certain period or within certain time ranges. A schematic of the vibration data interruption is shown in fig. 1. The vibration data shown in fig. 1 is interrupted between 8s and 11s, and the vibration amplitude is greatly reduced.
Based on the principle that the amplitude change of normal vibration data in the whole time period is stable, the effective value of the vibration data is calculated in a segmented mode through a sliding window, and whether the interruption abnormality exists in the vibration data is detected by judging whether the maximum effective value max _ r and the minimum effective value min _ r are different greatly. Suppose the vibration data sequence X ═ X1,x2,x3,…,xn]Wherein x isi(i-1 … n) is the data point in the vibration sequence and n is the data length. The method comprises the following specific steps:
1) setting the length N of a sliding window and a sliding step length s, wherein s is more than 0 and less than or equal to N; in order to avoid misjudging periodic impact in the vibration data as data interruption, the length of a data sliding window of each vibration measuring point needs to meet the condition: n is more than or equal to T multiplied by fs, wherein T is the rotating period of the rotating shaft corresponding to the vibration measuring point, and fs is the sampling frequency.
2) Sliding the sliding window backwards and successively by step length s from the starting time of the vibration data, and respectively calculating the effective value R of the vibration data in the sliding window after each sliding as [ R ═ R1,r2,r3,…,rK]Wherein, in the step (A),
(i ═ 0 … K, K is the number of window slips).
3) Min _ r ═ min (r) and max _ r ═ max (r) are calculated.
4) Computing
Comparing u with the data interruption threshold th when u is greater than>And th, judging that the vibration data is abnormal.
The value range of the vibration data interruption threshold th can be between 0.05 and 0.2, and the smaller the value of th is, the stricter the abnormity detection is.
Vibration data multiple-point repeat anomaly detection
Because the A/D bit number of the existing vibration data acquisition equipment is higher, the precision of converting a vibration analog signal into a digital signal is higher, and the number of repeated data in normal vibration data is less. In some cases, when the actual vibration exceeds the range of the sensor or the data acquisition equipment, the vibration data has a "clipping" phenomenon, and an abnormal phenomenon of repeated multiple points occurs in the vibration data. In addition, the fault of the edge data acquisition equipment can also cause the abnormal phenomenon of multi-point repetition in the vibration data. Accordingly, it is possible to identify whether the vibration data is abnormal by detecting whether there is a multi-point repetition in the vibration data. Suppose the vibration data sequence X ═ X1,x2,x3,…,xn]Wherein x isi(i-1 … n) is the data point in the vibration sequence and n is the data length. The vibration data multipoint repetition abnormity detection steps are as follows:
1) calculating a unique (non-repeating) sequence of values Y ═ Y in the vibration data X1,y2,y3,…,yk]. For example: x [ -1.2, -1.1, -1.25, -1.25,1.25,3,4,3]Then the unique numerical sequence Y of X [ -1.2, -1.1, -1.25,1.25,3,4]。
2) The number of times R _ num ═ m is calculated for each data point in Y that recurs in X1,m2,m3,…,mk]Wherein m isi(i-1 … k) for each data pointThe number of recurrences. The maximum value max _ num of the number of repeated occurrences of the repeated data point is calculated as max (R _ num).
3) And setting a repeated data point threshold th. In practice, th may take an integer value between 2-20, and the smaller the value of th, the more strict the anomaly detection is.
4) And comparing the size of max _ num with the size of th, and judging that the quality of the vibration data is abnormal when the size of max _ num is larger than th.
Tachometer pulse data anomaly detection
The rotating speed sensor is usually a proximity switch or an encoder, and in practice, the measured rotating speed pulse data is usually abnormal due to the problems of damage of the rotating speed sensor, unreasonable distance between the proximity switch and an induction mechanical element, loosening of the wiring of the rotating speed sensor and the like. The abnormal visual expression of the rotating speed pulse data is that the rotating speed pulse data has redundant interference pulses or loses due normal pulses. The rotational speed calculated from the rotational speed pulses cannot theoretically be higher than the limit rotational speed, and the rotational speed does not fluctuate greatly within a certain short period of time. The invention designs a rotating speed pulse data abnormity detection method based on the principle, which comprises the following specific steps:
1) extracting a time sequence t ═ t corresponding to the rising edge position of the rotating speed pulse data1,t2,,…,tn]；
2) Assuming that P pulses are generated in each rotation under normal conditions, sampling is carried out on T every P-1 points, and the time sequence T corresponding to the rising edge of the rotating speed pulse in the whole period is obtained as T1,T2,,…,Tn]；
3) Calculating the corresponding rotating speed sp of each whole periodi＝60/Ti(i＝1,2,…,n)；
4) Comparing sp one by oneiAnd the magnitude of the limit rotation speed (max _ r, min _ r) when sp isi>max _ r or spi<When min _ r is needed, judging that the rotating speed pulse data are abnormal; alternatively, the rotation speed fluctuation coefficient is calculated
If S>th, judging that the rotating speed pulse data is abnormal. Wherein the content of the first and second substances,th is a threshold value of the rotation speed fluctuation coefficient, the value of th can be between 0.01 and 0.05, and the smaller the value of th is, the stricter the abnormity detection is.
Detection of missing value abnormality in vibration data and rotation rate pulse data
And judging whether the value corresponding to each data point is a null value or not respectively point by point aiming at the vibration data and the rotating speed pulse data sequence uploaded by the edge end every time, and judging that the vibration data or the rotating speed pulse data are abnormal when one or more null values appear in the vibration data sequence and the rotating speed pulse data sequence. Here, the edge end refers to the position where the edge device is disposed, and the "edge" belongs to general terms in the field and is not described in detail.
Ratio of positive and negative data point number difference of vibration data
Because normal vibration data can fluctuate above and below 0, the number of positive and negative data points of the normal vibration data is close to that of the normal vibration data, and the proportion of the difference value of the positive and negative data points in the total data points is small. Based on the principle, the vibration data abnormity can be identified by judging the proportion of the positive data point and the negative data point of the vibration data. Suppose the vibration data sequence X ═ X1,x2,x3,…,xn]Wherein x isi(i-1 … n) is the data point in the vibration sequence and n is the data length. The detection steps of the abnormal ratio of the positive data points to the negative data points of the vibration data are as follows:
1) calculating vibration data points P1 greater than 0 and data points P2 less than 0 in the vibration data X;
2) calculating s ═ abs (P1-P2)/n, where abs (.) is the sign of the absolute value;
3) and setting a positive and negative data point number difference ratio threshold th. In practice, th can be a value between 0.001 and 0.1, and the smaller the value of th is, the stricter the abnormal detection is.
4) And comparing the sizes of s and th, and judging that the quality of the vibration data is abnormal when s is larger than th.
Vibration data and tachometer pulse data length anomaly detection
The vibration data and the rotating speed data of the monitoring system are acquired by edge data acquisition hardware, the length of the vibration data and the sampling rate thereof, the length of the rotating speed pulse data and the sampling rate thereof are preset, if the acquisition duration of the vibration data and the rotating speed pulse data is t & lt20 s & gt, the sampling rate is fs & lt12800 Hz, and the length of the vibration data and the rotating speed pulse data uploaded by the edge end every time is N & ltfs & gt × t & lt 256000. The method specifically comprises the following steps of detecting abnormal lengths of vibration data and rotating speed pulse data:
1) calculating the lengths of vibration data and rotating speed pulse data of each time of uploading data by the edge end as vib _ n and imp _ n respectively;
2) judging whether the vib _ n is equal to the preset data length or not, and if not, judging that the vibration data quality is abnormal;
and judging whether the imp _ n is equal to the preset data length or not, and if not, judging that the quality of the rotating speed pulse data is abnormal.
Vibration data over-sensor range anomaly detection
When the vibration sensor is damaged or the edge data acquisition equipment is interfered by external electromagnetic waves, the abnormal phenomenon that the vibration data exceed the measuring range of the sensor can occur. The vibration acceleration sensors commonly used in industrial equipment such as wind turbines are classified into low frequency and normal frequency, and the measuring range of the normal frequency vibration acceleration sensor is usually 50g or 80 g. The low frequency vibration acceleration sensor typically has a range of 10 g. Suppose the vibration data sequence X ═ X1,x2,x3,…,xn]Wherein x isi(i-1 … n) is the data point in the vibration sequence and n is the data length. The specific steps for judging whether the vibration data exceeds the range of the sensor are as follows:
1) calculating a peak value P _ P ═ max (X) -min (X) of each time the edge data acquisition equipment uploads the vibration data;
2) and comparing the magnitude of P _ P and 2A (A is the sensor range), and if P _ P is greater than 2A, judging that the vibration data exceeds the sensor range.
Examples
The present embodiment introduces embodiments of detecting multipoint repeated abnormality of vibration data, detecting drifting abnormality of vibration data, detecting interruption abnormality of vibration data, detecting singular value abnormality of vibration data, and detecting abnormality of rotational speed pulse data, which are related in the present invention, by using simulation data, respectively.
(1) Vibration data multiple-point repeat anomaly detection
A set of 20s of multi-point repetitive vibration data is generated at a sampling frequency fs of 12800Hz, which is shown in fig. 2. In fig. 2, the vibration data has a multi-point data repeat phenomenon between 10s and 13s, and the maximum value max _ num of the number of repeated occurrences of the repeated data points is calculated to be 19200. And (5) judging that the vibration data is abnormal by taking 10 as the multipoint repetition threshold th of the vibration data and judging that max _ num is far larger than th.
(2) Vibration data drift anomaly detection
A set of 20s vibration data drift simulation waveforms is generated at a sampling frequency fs of 12800Hz, as shown in fig. 3. Setting the length N of the sliding window to be 12800; the step length s of sliding is 12800, and the sliding times are 20 times. The maximum value Ma of the absolute value of the mean of the vibration data in 20 sliding windows was calculated as 9.8519. Threshold th is 0.4m/s2；Ma>th, it is judged that the vibration data is abnormal.
(3) Vibration data interrupt anomaly detection
A set of 20s vibration data-interrupted simulation waveforms is generated at a sampling frequency fs of 12800Hz, as shown in fig. 4. Setting the length N of a sliding window to be 25600, the sliding step length s to be 12800, calculating effective values of vibration data in the window in a sliding mode, and calculating the effective value R of the vibration data in each window to be R1,r2,r3,…,r19]And calculating the maximum value max _ R of R as 20.119m/s2Minimum min _ r is 1.003m/s2。
The value of the abnormal detection threshold th of the vibration data interruption is usually between 0.05 and 0.2, and u is>th, it is judged that the vibration data is abnormal.
(4) Singular value anomaly detection for vibration data
At a sampling frequency fs of 12800Hz, a set of 20s vibration data having a plurality of singular values is generated, and the waveform thereof is as shown in fig. 5. The root mean square sd of the vibration data shown in fig. 5 was calculated to be 3.664m/s2Taking the root mean square number k as 6, and calculating the originalThe absolute value of the vibration data is greater than 6 × sd, and the ratio P of the number of data points (singular value points) to the total data amount is 20, i.e., 0.000078. Assume that singular value abnormality detection threshold th is 0.00006, P>th, judging that the vibration data is abnormal.
(5) Tachometer pulse data anomaly detection
Taking a wind turbine generator as an example, the sampling frequency fs is 12800Hz, a group of 20s of abnormal data of the rotating speed pulse of the high-speed shaft of the gearbox interfered by the redundant pulse is generated, and the waveform of the abnormal rotating speed pulse data is shown in fig. 6. If the limit speed is 1800rpm, 1 pulse is generated per revolution of the high speed shaft of the gearbox. The method according to the invention calculates the spin sp over the entire period of time from the rising edges of two adjacent spin pulses, the calculation result being shown in fig. 7. As can be seen from fig. 7, a plurality of data exceeding the limit rotational speed appear in the rotational speed information over the entire period, whereby it is determined that the rotational speed pulse data is abnormal.
Calculating the coefficient of speed fluctuation in FIG. 7 according to the method of the present invention
The threshold th of the fluctuation coefficient of the rotation speed is 0.04, s>th, also indicates the tachometer pulse data anomaly.
Finally, it should be noted that the present application is only an example of an industrial device, and those skilled in the art will understand that data anomaly detection in other fields is also applicable.
Although the embodiments disclosed in the present application are described above, the descriptions are only for the convenience of understanding the present application, and are not intended to limit the present application. It will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the disclosure as defined by the appended claims.
Claims (6)
1. The data quality abnormity detection method of the monitoring system is characterized by comprising the steps of detecting the drifting abnormity of vibration data and detecting the interruption abnormity of the vibration dataAt least one, obtaining vibration data sequence by arranging vibration sensorX=[x 1,x 2,x 3, …,x n ]Whereinx i（i=1…n) Are the data points in the vibration sequence and,nis the data length; the abnormal drift detection of the vibration data adopts a sliding window to calculate the mean value of the vibration data in a segmentation way, and the maximum value of the absolute value of the mean value of each segment of the vibration data is takenM aBy judgmentM aWhether the vibration data deviate from the 0 line is identified;
the step of detecting the abnormal drifting of the vibration data is as follows:
setting the length of the first sliding windowN 1And a first sliding steps 1Wherein, 0<s 1≤N 1；
Starting the sliding window with step size from the starting time of the vibration datas 1Sliding backwards successively, and calculating the absolute value of the mean value of the vibration data in the sliding window after each slidingM=[m 1,m 2,m 3, …,m K ]Wherein, in the step (A),
（ i=0…K，Kwindow sliding times);
computingMMaximum value ofM a=max(M)；
4) ComparisonM aAnd a predetermined vibration data drift threshold thMaThe size of (1) whenM a>thMaJudging that the quality of the vibration data is abnormal;
the vibration data interruption anomaly detection adopts a sliding window to calculate the effective value of the vibration data in a segmented mode, and whether interruption anomaly exists in the vibration data is detected by judging whether the maximum effective value max _ r and the minimum effective value min _ r are greatly different; the step of detecting the abnormal interruption of the vibration data is as follows:
setting the length of the second sliding windowN 2And a second sliding steps 2Wherein, 0<s 2≤N 2(ii) a The length of the data sliding window of each vibration measuring point meets the condition:N 2≥T×fswherein, in the step (A),Tin order to correspond to the rotation period of the rotating shaft of the vibration measuring point,fsis the sampling frequency;
starting the sliding window with step size from the starting time of the vibration datas 2Sliding backwards successively, and calculating the effective value of vibration data in the sliding window after each slidingR=[r 1,r 2,r 3, …,r K ]Wherein, in the step (A),
（ i=0…K，Kwindow sliding times);
calculate min _ r = min (R)，max_r=max(R)；
Computing
Comparison ofuAnd a predetermined data interruption threshold th u The size of (1) whenu>th u Judging that the vibration data is abnormal;
the method also comprises the step of detecting the multipoint repetition abnormity of the vibration data, wherein the multipoint repetition abnormity of the vibration data comprises the following steps:
calculating vibration dataXIn a non-repeating unique sequence of valuesY=[y 1,y 2,y 3, …,y k ]；
ComputingYEach data point inXIn the number of repeated occurrencesR_num=[m 1,m 2,m 3, …,m k ]Whereinm i （i=1…k) For the number of repetitions of each data point, the maximum number of repetitions of the repeated data point max _ num = max (m:)R_num)；
3) Setting a repeated data point threshold th1, comparing the sizes of max _ num and th1, and judging that the quality of the vibration data is abnormal when max _ num is greater than th 1;
the method also comprises the abnormal detection of the rotating speed pulse data, wherein the rotating speed pulse data is obtained by arranging a rotating speed sensor, and the abnormal detection of the rotating speed pulse data comprises the following specific steps:
extracting time sequence corresponding to rising edge position of rotating speed pulse datat=[t 1,t 2, …,t n]；
Assuming normal conditions producing each revolutionP 0Every pulse, thenP 0-1 point pairtSampling to obtain the time sequence corresponding to the rising edge of the whole-period rotating speed pulseT=[T 1, T 2 ,…,T n ]；
Calculating the corresponding rotating speed of each whole periodsp i=60/T i （i=1,2, …,n）；
4) Compare one by onesp i With the maximum and minimum rotational speeds max _ r and min _ r, respectivelysp i >max _ r orsp i <When min _ r is needed, judging that the rotating speed pulse data are abnormal; alternatively, the rotation speed fluctuation coefficient is calculated
If, ifs 0>th2, judging the abnormal data of the rotating speed pulse; th2 is a preset threshold value of the fluctuation coefficient of the rotation speed.
2. The method for detecting the abnormal data quality of the monitoring system according to claim 1, further comprising the step of detecting the abnormal singular value of the vibration data, wherein the step of detecting the abnormal singular value of the vibration data comprises the following steps:
computingXRoot mean square (rms):
；
computingX 1=abs(X)；
Setting root mean square coefficientkK has a value between 3 and 8;
statistics ofX 1Is greater thankNumber of data num of xsd and calculating singular value ratio P = num-n；
Setting singular value anomaly detection threshold thPComparing P and thPSize when P>thPAnd judging that the quality of the vibration data is abnormal.
3. The method for detecting the data quality abnormality of the monitoring system according to the claim 1 or 2, characterized by further comprising the step of detecting the positive and negative data point difference ratio of the vibration data, wherein the detecting step is as follows:
calculating vibration dataXThe vibration data point number P1 greater than 0 and the data point number P2 less than 0;
computings 3=abs(P1-P2)/n；
Setting a positive and negative data point number difference ratio threshold th 3;
comparisons 3And th3 size whens 3>th3, it is judged that the vibration data quality is abnormal.
4. The method for detecting data quality abnormality of a monitoring system according to claim 3, further comprising detecting abnormal range of vibration data over a sensor, the detecting steps are as follows:
calculating the peak-to-peak value P _ P = max of the vibration data uploaded by the edge data acquisition equipment each time (bX)-min(X)；
Compare P _ P and 2AThe size of (a) is (b),Ais the sensor range; if P _ P>2AAnd judging that the vibration data exceeds the range of the sensor.
5. The method for detecting the data quality abnormality of the monitoring system according to the claim 1, the claim 2 or the claim 4, further comprising the step of detecting the abnormality of the vibration data and the length of the rotating speed pulse data, and the step comprises the following steps:
calculating the lengths of vibration data and rotating speed pulse data of each time of uploading data by the edge end as vib _ n and imp _ n respectively;
judging whether the vib _ n is equal to the preset data length or not, and if not, judging that the vibration data quality is abnormal;
and judging whether the imp _ n is equal to the preset data length or not, and if not, judging that the quality of the rotating speed pulse data is abnormal.
6. The method for detecting the abnormal data quality of the monitoring system according to claim 5, further comprising abnormal detection of missing values of the vibration data and the rotation speed pulse data, namely judging whether a value corresponding to each data point is a 'null value' point by point respectively aiming at the vibration data and the rotation speed pulse data sequence uploaded at each time of the edge end, and judging that the vibration data or the rotation speed pulse data are abnormal when one or more null values appear in the vibration data sequence and the rotation speed pulse data sequence.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910984189.5A CN110674891B (en) | 2019-10-16 | 2019-10-16 | Data quality abnormity detection method for monitoring system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910984189.5A CN110674891B (en) | 2019-10-16 | 2019-10-16 | Data quality abnormity detection method for monitoring system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110674891A CN110674891A (en) | 2020-01-10 |
| CN110674891B true CN110674891B (en) | 2021-11-30 |
Family
ID=69082715
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910984189.5A Active CN110674891B (en) | 2019-10-16 | 2019-10-16 | Data quality abnormity detection method for monitoring system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110674891B (en) |
Families Citing this family (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111711678A (en) * | 2020-06-09 | 2020-09-25 | 重庆川仪自动化股份有限公司 | Large-scale mobile equipment state monitoring system and method based on Internet of things technology |
| CN114660323B (en) * | 2020-12-22 | 2023-07-18 | 莱克电气股份有限公司 | Rotation speed measuring method, rotation speed measuring device, rotation speed measuring equipment and storage medium |
| CN113377842B (en) * | 2021-06-21 | 2022-07-08 | 山东八五信息技术有限公司 | Emergency management method and system based on industrial internet big data |
| CN113688791B (en) * | 2021-09-22 | 2024-03-01 | 西安热工研究院有限公司 | Method for identifying CMS abnormal data packet of wind turbine generator |
| CN115900928A (en) * | 2021-09-30 | 2023-04-04 | 深圳富桂精密工业有限公司 | Data acquisition system |
| CN114217119B (en) * | 2021-12-07 | 2023-12-19 | 广西电网有限责任公司电力科学研究院 | Data distortion identification method and system based on sampling values at different moments |
| CN115017121B (en) * | 2022-08-05 | 2022-10-25 | 山东天意机械股份有限公司 | Data storage system of concrete production equipment |
| CN117971625B (en) * | 2024-03-27 | 2024-06-07 | 莱芜职业技术学院 | Performance data intelligent monitoring system based on computer cloud platform |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1707076A (en) * | 2004-06-09 | 2005-12-14 | 五十铃自动车株式会社 | Fatigue failure diagnostic method of turbocharger and fatigue failure diagnostic apparatus for turbocharger |
| CN102840882A (en) * | 2012-09-04 | 2012-12-26 | 中国海洋石油总公司 | State monitoring and fault diagnosis system of gas turbine generating unit and use method of state monitoring and fault diagnosis system |
| CN104111157A (en) * | 2014-07-14 | 2014-10-22 | 中国空气动力研究与发展中心高速空气动力研究所 | Method for examining initial readings of wind tunnel measurement instrument |
| CN105043770A (en) * | 2015-05-13 | 2015-11-11 | 北京天源科创风电技术有限责任公司 | Wind turbine generator abnormal vibration judging method and apparatus thereof |
| CN107064796A (en) * | 2017-04-20 | 2017-08-18 | 哈尔滨理工大学 | Intelligent electric machine vibration online monitoring and early warning system |
| CN107367687A (en) * | 2017-07-31 | 2017-11-21 | 江苏省电力试验研究院有限公司 | A kind of high-voltage breaker spring fault degree detection method and device |
| CN107966307A (en) * | 2017-12-14 | 2018-04-27 | 常州路航轨道交通科技有限公司 | A kind of train operating safety on-line monitoring method based on vibration acceleration |
| CN109543357A (en) * | 2018-10-18 | 2019-03-29 | 电子科技大学 | Fault degree quantitative evaluation method for multivariate regression model optimization |
| CN109682601A (en) * | 2019-03-04 | 2019-04-26 | 北京天泽智云科技有限公司 | The initial failure recognition methods of rolling bearing under a kind of variable speed operating condition |
| CN109909803A (en) * | 2019-04-17 | 2019-06-21 | 北京天泽智云科技有限公司 | A kind of machine tool chief axis method for detecting abnormality |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| GB2491804B (en) * | 2011-05-11 | 2018-01-17 | Syrinix Ltd | Pipeline fault detection system and monitor unit |
-
2019
- 2019-10-16 CN CN201910984189.5A patent/CN110674891B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1707076A (en) * | 2004-06-09 | 2005-12-14 | 五十铃自动车株式会社 | Fatigue failure diagnostic method of turbocharger and fatigue failure diagnostic apparatus for turbocharger |
| CN102840882A (en) * | 2012-09-04 | 2012-12-26 | 中国海洋石油总公司 | State monitoring and fault diagnosis system of gas turbine generating unit and use method of state monitoring and fault diagnosis system |
| CN104111157A (en) * | 2014-07-14 | 2014-10-22 | 中国空气动力研究与发展中心高速空气动力研究所 | Method for examining initial readings of wind tunnel measurement instrument |
| CN105043770A (en) * | 2015-05-13 | 2015-11-11 | 北京天源科创风电技术有限责任公司 | Wind turbine generator abnormal vibration judging method and apparatus thereof |
| CN107064796A (en) * | 2017-04-20 | 2017-08-18 | 哈尔滨理工大学 | Intelligent electric machine vibration online monitoring and early warning system |
| CN107367687A (en) * | 2017-07-31 | 2017-11-21 | 江苏省电力试验研究院有限公司 | A kind of high-voltage breaker spring fault degree detection method and device |
| CN107966307A (en) * | 2017-12-14 | 2018-04-27 | 常州路航轨道交通科技有限公司 | A kind of train operating safety on-line monitoring method based on vibration acceleration |
| CN109543357A (en) * | 2018-10-18 | 2019-03-29 | 电子科技大学 | Fault degree quantitative evaluation method for multivariate regression model optimization |
| CN109682601A (en) * | 2019-03-04 | 2019-04-26 | 北京天泽智云科技有限公司 | The initial failure recognition methods of rolling bearing under a kind of variable speed operating condition |
| CN109909803A (en) * | 2019-04-17 | 2019-06-21 | 北京天泽智云科技有限公司 | A kind of machine tool chief axis method for detecting abnormality |
Non-Patent Citations (2)
| Title |
|---|
| 振动测试异常值的原因分析;贾建峰 等;《计测技术》;20190317;第39卷(第3期);全文 * |
| 风电机组振动监测与故障诊断系统研究;曹斌;《中国优秀硕士学位论文全文数据库 工程科技Ⅱ辑》;20141015(第10期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110674891A (en) | 2020-01-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110674891B (en) | Data quality abnormity detection method for monitoring system | |
| Sheng | Wind turbine gearbox condition monitoring round robin study-vibration analysis | |
| Villa et al. | Statistical fault diagnosis based on vibration analysis for gear test-bench under non-stationary conditions of speed and load | |
| Shahriar et al. | Electrical signature analysis-based detection of external bearing faults in electromechanical drivetrains | |
| EP2026160B1 (en) | Event monitoring via combination of signals | |
| Bechhoefer et al. | Processing for improved spectral analysis | |
| Jablonski et al. | Validation of vibration measurements for heavy duty machinery diagnostics | |
| CN107003663A (en) | The monitoring of device with movable part | |
| CN111946559B (en) | Method for detecting structures of wind turbine foundation and tower | |
| CN115935243B (en) | Fault analysis method based on data processing | |
| CN105424333B (en) | A kind of monitoring of pneumatic equipment bladess on-site damage and recognition methods | |
| CN102352824B (en) | Monitoring system based on electric information for health status of wind driven generator and monitoring method thereof | |
| Lin et al. | A review and strategy for the diagnosis of speed-varying machinery | |
| Kannan et al. | Demodulation band optimization in envelope analysis for fault diagnosis of rolling element bearings using a real-coded genetic algorithm | |
| CN112017409A (en) | Trend early warning method for short-time increase amplitude of mechanical equipment vibration | |
| KR102545672B1 (en) | Method and apparatus for machine fault diagnosis | |
| CN112177865A (en) | Method for solving marking noise and insufficient marks in fan fault detection | |
| CN113280910A (en) | Real-time monitoring method and system for long product production line equipment | |
| KR102171505B1 (en) | body of revolution of teeth based rotate-body defect analysis apparatus and method thereof | |
| CN112347944A (en) | Mechanical transmission fault detection method based on zero-value proportional spectrum characteristics | |
| CN113688791B (en) | Method for identifying CMS abnormal data packet of wind turbine generator | |
| CN116499742A (en) | Rotary machine fault monitoring method based on ensemble learning | |
| CN113218653B (en) | Offshore wind power gear box monitoring method and device based on low-frequency disturbance elimination | |
| Song et al. | Framework of designing an adaptive and multi-regime prognostics and health management for wind turbine reliability and efficiency improvement | |
| JP2003344150A (en) | Condition diagnosing method and apparatus therefor by analyzing shaft vibration of rotating machine |
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 |