A time domain calculation method for processing steam turbine blade vibration signal trigger time

Abstract

A kind of time domain calculation method for processing steam turbine blade vibration signal trigger time, belong to steam turbine detection technical field.To solve the problem of inaccuracy of trigger time calculation of non-uniform steam turbine blade, the field blade vibration monitoring system collects data, including time series data and signal column data;Using two hampel filters to remove outliers of the collected signal column data, using a fixed data interval to resample the signal column data after removing outliers, performing multiple smoothing filter processing on the obtained resampled signal column data, performing ascending and descending sequence calibration on the obtained smoothed signal column data after filtering, finding the maximum and minimum values, and identifying the upper envelope and lower envelope;The obtained envelope is multiplied by a coefficient as the trigger line of the signal, and the intersection point of the obtained smoothed signal column data after filtering and the trigger line is taken as the steam turbine blade vibration signal trigger time.The present application has good applicability.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of steam turbine detection, and particularly relates to a time domain calculation method for processing a triggering time of a steam turbine blade vibration signal. BACKGROUND

[0002] A steam turbine is a rotary machine, and at present, various steam turbine manufacturers install non-contact distance sensors at the radial positions of the last-stage blades of the steam turbine to achieve the purpose of monitoring the vibration of the steam turbine blades. Since there are gaps between the steam turbine blades, the collected signals are periodic signals, and an important data in the algorithm for monitoring the steam turbine blades is the triggering time, that is, the time when each blade passes through the sensor.

[0003] The conventional method for calculating the triggering time of the steam turbine blade is to set a fixed triggering threshold, and the time when the signal coincides with the threshold is taken as the passing time of the blade. However, due to the large fluctuation and non-smoothness of the actual signal, two time points are triggered at the adjacent two points, in order to avoid the above problems, a frequency domain filter is added to the conventional calculation method, and the signal passes through one or more filters to make the data more smooth, and the triggering time is calculated. However, the conventional calculation method has three problems in identifying the triggering time. The first problem is that when using the frequency domain method to filter, the waveform after the combined filter is very different from the original waveform, resulting in inaccurate calculation of the triggering time. The second problem is that the triggering threshold is a fixed value, and when there is a small fluctuation in the original signal, the calculated triggering time is inaccurate. The third problem is that due to the problem of non-uniformity of the installation of the steam turbine blades, the distance of some blades to the sensor is inconsistent with that of other blades, which further increases the signal fluctuation. In this case, the blade signal and the triggering threshold are completely not intersected, resulting in the loss of the triggering time and affecting the realization of the subsequent functions. The first two problems may have some deviation in the calculation time, but they can be identified by the conventional method. However, if problem three occurs, the conventional method cannot meet the identification requirements. SUMMARY

[0004] The present application solves the problem of inaccurate calculation of the triggering time of the non-uniform steam turbine blades, and proposes a time domain calculation method for processing the triggering time of the steam turbine blade vibration signal.

[0005] To achieve the above-mentioned purpose, the present application realizes the following technical scheme:

[0006] A time domain calculation method for processing the triggering time of a steam turbine blade vibration signal, comprising the following steps:

[0007] S1, a field blade vibration monitoring system collects data, including time column data and signal column data;

[0008] S2. Use the Hampel filter twice to remove outliers from the signal train data acquired in step S1, and obtain the signal train data after removing outliers;

[0009] S3. Resample the signal train data after removing outliers in step S2 using a fixed data interval to obtain the resampled time train data and signal train data.

[0010] S4. Perform multiple smoothing filtering processes on the resampled signal train data obtained in step S3 to obtain smoothed signal train data.

[0011] S5. Perform ascending and descending order calibration on the signal train data after smoothing and filtering obtained in step S4, find the maximum and minimum values, and identify the upper envelope and lower envelope.

[0012] S6. Multiply the envelope obtained in step S5 by a coefficient to obtain the signal trigger line. The intersection of the signal train data after smoothing and filtering obtained in step S4 with the trigger line is taken as the trigger time of the turbine blade vibration signal.

[0013] In the further step S1, the method for the on-site blade vibration monitoring system to collect data is to acquire on-site blade vibration data through a magnetic displacement sensor, and then store the data on a local server via a digital acquisition unit. The formulas for the obtained time series data and signal series data are as follows:

[0014] T = (t1, t2, ... t) n-1 ,t n )

[0015] Y = (y1, y2, ..., y3) n-1 ,y n )

[0016] Where T represents the time series data, Y represents the signal series data, t1, t2...t n-1 ,t n These are time points, y1, y2...y n-1 ,y n These are the signal data at the corresponding time points, where n is the number of data points.

[0017] Furthermore, the specific implementation method of step S2 includes the following steps:

[0018] S2.1 Set the parameters of the Hampel filter, including the window size K and the standard deviation coefficient N. σ ;

[0019] S2.2, Regarding t collected in step S1 i Signal data y at time t i(i∈1,2...n), uniformly take y i K values of Y near both sides i , Y i is expressed as Calculate the median m i and the standard deviation σ i of Y i ;

[0020] S2.3, hampel filter filtering is performed on the signal data y i , and the filtered signal data is If

[0021] |y i -m i |>N σ ×σ i , Take the median m i , otherwise maintain the original signal data unchanged, and the calculation formula of the hampel filter filtering is:

[0022]

[0023] S2.4, the parameters of the hampel filter are set again, and steps S2.2-S2.3 are repeated to perform the second hampel filter filtering, and after 2 times of hampel filtering, the signal data is

[0024] Further, the value range of K in the first hampel filtering in step S2.1 is 400-500, the value range of N σ is 2-3, the value range of K in the second hampel filtering in step S2.4 is 40-50, and the value range of N σ is 0.5-1.

[0025] Further, the value range of the data interval k c of data resampling in step S3 is 100-150, the resampled data is h, and the time column data T c and the signal column data Y c after resampling are as follows:

[0026]

[0027]

[0028] Among them, are the resampling time points, are the resampled signal data corresponding to the time points, and h is the number of data.

[0029] Furthermore, the smoothing filtering process in step S4 involves applying a mean filter to the resampled signal data, using a five-point cubic smoothing filter method. The calculation formula is as follows:

[0030]

[0031]

[0032]

[0033]

[0034]

[0035] in, The signal data is filtered; the smoothing filtering process is performed 8-10 times. The filtered signal data has good smoothness, and the signal data of one blade cycle has a peak and a trough value.

[0036] Furthermore, the specific implementation method of step S5 includes the following steps:

[0037] S5.1, Filtered signal data The data in the middle is obtained by first subtracting the two sets of data. The calculation formula is as follows:

[0038]

[0039] S5.2, then Y d Calculate the ascending / descending order calibration signal data. Where Y d If the difference is greater than 0, then Y s For a value calibrated as 1, Y d The difference equals 0Y s For a value calibrated to 0, Y d The difference is less than 0, Y s The value is calibrated as -1;

[0040] S5.3 Repeat step S5.2 and record Y. d The time when the value of is equal to 2, and the corresponding Y at that time. b The value of is the intersection point of the envelope on the curve, and Y is recorded. d The time at which the value equals -2, and the corresponding Y at that time. b The value of is the intersection point of the lower envelope of the curve. Plot the intersection points of the upper and lower envelopes on the curve to obtain the upper and lower envelope lines.

[0041] Furthermore, the coefficient in step S6 is the scaling factor k1, and the value of k1 is 0.7.

[0042] Advantages of the present application:

[0043] The time domain calculation method for processing the triggering time of the steam turbine blade vibration signal provided by the present application adopts the time domain method to process the signal, uses two-layer hampel filters to eliminate the outliers of the signal for the signal collected by the field sensor, resamples the processed data to eliminate the small-range fluctuation, uses the smoothing filter to process the data to make the signal locally monotonic, and calculates the upper and lower envelope lines of the processed data, and multiplies the envelope lines by a coefficient as the signal triggering threshold, so that the method has good applicability to the data with large fluctuations. BRIEF DESCRIPTION OF DRAWINGS

[0044] Figure 1 The process flow chart of the time domain calculation method for processing the triggering time of the steam turbine blade vibration signal provided by the present application;

[0045] Figure 2 The time domain signal diagram collected on site for the time domain calculation method for processing the triggering time of the steam turbine blade vibration signal provided by the present application;

[0046] Figure 3 The time domain signal diagram after filtering outliers for the time domain calculation method for processing the triggering time of the steam turbine blade vibration signal provided by the present application;

[0047] Figure 4 The time domain signal diagram after smoothing filter processing for the time domain calculation method for processing the triggering time of the steam turbine blade vibration signal provided by the present application;

[0048] Figure 5 The time domain signal diagram with upper and lower envelope lines for the time domain calculation method for processing the triggering time of the steam turbine blade vibration signal provided by the present application;

[0049] Figure 6 The time domain signal diagram with the triggering time of the steam turbine blade vibration signal for the time domain calculation method for processing the triggering time of the steam turbine blade vibration signal provided by the present application. DETAILED DESCRIPTION

[0050] In order to make the purpose, technical solutions and advantages of the present application clearer and more apparent, the present application is further described in detail below in combination with the drawings and specific embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application, that is, the described specific embodiments are only a part of the embodiments of the present application, but not all the specific embodiments. The components of the specific embodiments of the present application described and shown in the drawings herein can be arranged and designed in various different configurations, and the present application can also have other embodiments.

[0051] Therefore, the following detailed description of the specific embodiments of the application provided in the drawings is not intended to limit the scope of the claimed application, but merely represents selected specific embodiments of the application. Based on the specific embodiments of the application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of the protection of the application.

[0052] In order to further understand the invention content, characteristics and effects of the present application, the following specific embodiments are exemplified, and the accompanying drawings are provided Figure 1 -attached Figure 6 The detailed description is as follows: Specific embodiment one:

[0054] A time domain calculation method for processing steam turbine blade vibration signal triggering time, comprising the following steps:

[0055] S1, the field blade vibration monitoring system collects data, including time column data and signal column data;

[0056] Further, the method for collecting data by the field blade vibration monitoring system in step S1 is to collect field blade vibration data through a magnetic pull type displacement sensor, and then store the time column data and signal column data obtained through a digital collector into a local server. The formula is:

[0057] T=(t1,t2...t n-1 ,t n )

[0058] Y=(y1,y2...y n-1 ,y n )

[0059] Wherein, T is the time column data, Y is the signal column data, t1, t2...t n-1 ,t n are respectively the time, y1, y2...y n-1 ,y n are respectively the signal data corresponding to the time, and n is the number of data;

[0060] S2, using two times hampel filter to remove the wild point of the signal column data collected in step S1, to obtain the signal column data after removing the wild point;

[0061] Further, the specific implementation method of step S2 includes the following steps:

[0062] S2.1, setting the parameters of the hampel filter, the parameters of the hampel filter including window size K and standard deviation coefficient N σ ;

[0063] Furthermore, in step S2.1, the value of K for the first Hampel filter ranges from 400 to 500, and N... σ The value range is 2-3;

[0064] S2.2, Regarding t collected in step S1 i Signal data y at time t i (i∈1,2...n), uniformly select y i K numerical groups Y on both sides i Y i Represented as Calculate Y i median m i and standard deviation σ i ;

[0065] S2.3, Regarding signal data y i Perform Hampshire filter filtering, the filtered signal data is like

[0066] |y i -m i |>N σ ×σ i hour, Take the median m i Otherwise, the original signal data remains unchanged. The calculation formula for the Hampel filter is:

[0067]

[0068] S2.4. Reset the parameters of the Hampel filter and repeat steps S2.2-S2.3 to perform a second Hampel filter pass. After two Hampel filters, the signal data is...

[0069] Furthermore, in step S2.4, the value of K for the second Hampel filter ranges from 40 to 50, and N... σ The value range is 0.5-1;

[0070] S3. Resample the signal train data after removing outliers in step S2 using a fixed data interval to obtain the resampled time train data and signal train data.

[0071] Furthermore, the data interval k for data resampling in step S3 c The value range is 100-150, the resampled data consists of h data points, and the resampled time series data is T. c and signal column data Y c It is expressed as follows:

[0072]

[0073]

[0074] in, These are the resampling times. These are the resampled signal data at the corresponding time points, where h is the number of data points;

[0075] S4. Perform multiple smoothing filtering processes on the resampled signal train data obtained in step S3 to obtain smoothed signal train data.

[0076] Furthermore, the smoothing filtering process in step S4 involves applying a mean filter to the resampled signal data, using a five-point cubic smoothing filter method. The calculation formula is as follows:

[0077]

[0078]

[0079]

[0080]

[0081]

[0082] in, The signal data is filtered; the smoothing filtering process is performed 8-10 times. The filtered signal data has good smoothness. The signal data of one blade cycle has a peak and a trough value.

[0083] S5. Perform ascending and descending order calibration on the signal train data after smoothing and filtering obtained in step S4, find the maximum and minimum values, and identify the upper envelope and lower envelope.

[0084] Furthermore, the specific implementation method of step S5 includes the following steps:

[0085] S5.1, Filtered signal data The data in the middle is obtained by first subtracting the two sets of data. The calculation formula is as follows:

[0086]

[0087] S5.2, then Y d Calculate the ascending / descending order calibration signal data. Where Y d If the difference is greater than 0, then Y s For a value calibrated as 1, Y d The difference equals 0Ys For the value of Y d Difference value is less than 0, Y s For the value of Y

[0088] S5.3, repeat step S5.2, record Y d The time point of the point whose value of Y b The value of Y d The time point of the point whose value is -2, and Y b The value of Y

[0089] S6, multiply the envelope line obtained in step S5 by a coefficient as a trigger line of the signal, and the intersection point of the signal column data obtained by the smoothing filtering in step S4 and the trigger line is taken as the trigger time point of the steam turbine blade vibration signal.

[0090] Further, the coefficient of step S6 is a scaling coefficient k1, and the value of k1 is 0.7.

[0091] It should be noted that the terms "first", "second", and so on, and the like relational terms are merely used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between the entities or operations. Moreover, the terms "include", "contain" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not explicitly listed or inherent to such process, method, article or device. Without more limitations, the element defined by the statement "including a" does not exclude the presence of another identical element in the process, method, article or device including the element.

[0092] Although the present application has been described above with reference to specific embodiments, various modifications can be made thereto and equivalents can be substituted for elements thereof without departing from the scope of the present application. In particular, features of the specific embodiments disclosed herein can be combined together in any manner without structural conflicts, and the combinations are not exhaustively described in the specification only for the purpose of omitting the length and saving resources. Therefore, the present application is not limited to the specific embodiments disclosed herein, but includes all technical solutions falling within the scope of the claims.

Claims

1. A time domain calculation method for processing the trigger time of a vibration signal of a steam turbine blade, characterized in that: It comprises the following steps: S1, the field blade vibration monitoring system collects data, including time series data and signal column data; S2, using twice hampel filter to remove the wild point of signal column data collected in step S1, to obtain the signal column data after removing wild points; S3, using fixed data interval to resample the signal column data after removing wild points in step S2, to obtain the resampled time series data and signal column data; S4, multiple smoothing filter processing is performed on the resampled signal column data obtained in step S3, to obtain the signal column data after smoothing filter processing; S5, ascending and descending sequence calibration is performed on the signal column data after smoothing filter processing obtained in step S4, to obtain the maximum value and minimum value, and to identify the upper envelope line and the lower envelope line; S6, multiplying the envelope line obtained in step S5 by a coefficient as the trigger line of the signal, and taking the intersection point of the signal column data after smoothing filter processing obtained in step S4 and the trigger line as the steam turbine blade vibration signal triggering time.

2. The time domain calculation method for processing the vibration signal trigger time of a steam turbine blade according to claim 1, characterized in that: The method for collecting data by the field blade vibration monitoring system in step S1 is to collect field blade vibration data through a magnetic pull displacement sensor, and then store the data in a local server through a digital collector, to obtain the time series data and signal column data, and the formula is: T = (t1, t2...t n-1 , n ) Y = (yl, y2... y n-1 ,y n ) Wherein, T is time column data, Y is signal column data, t1, t2...t n-1 , n are time, y1, y2...y n-1 , n are signal data corresponding to the time, and n is the number of data.

3. The time domain calculation method for processing the triggering time of the vibration signal of the turbine blade according to claim 1 or 2, characterized in that: The specific implementation method of step S2 comprises the following steps: S2.1, setting parameters of a hampel filter, the parameters of the hampel filter including a window size K and a standard deviation coefficient N σ ; S2.2, for the signal data y i collected in step S1 i (i∈1,2...n), uniform y i two sides near K value group Y i , Y i is expressed as the median m i and the standard deviation σ i of Y i ; S2.3, signal data y i hampel filter, the filtered signal data is If |y i -m i |>N σ ×σ i , Take the median m i , otherwise the original signal data is maintained unchanged, the calculation formula of hampel filter is: S2.4, the parameters of the Hampel filter are set again, steps S2.2-S2.3 are repeated, and the second Hampel filter filtering is performed. After 2 times of Hampel filtering, the signal data is 4. The time domain calculation method for processing the vibration signal trigger time of a steam turbine blade according to claim 3, characterized in that: The value range of K in the first time of Hampel filtering in step S2.1 is 400-500, N σ The value range of K in the first time of Hampel filtering in step S2.1 is 400-500, N σ The value range of K in the first time of Hampel filtering in step S2.1 is 400-500, N 5. The time domain calculation method for processing the vibration signal trigger time of a steam turbine blade according to claim 4, characterized in that: Step S3 data resampling data interval k c The value range of k is 100-150, the resampled data is h, and the resampled time column data T c And signal column data Y c Is expressed as follows: wherein, are the resampling instants, respectively, are the resampled signal data at the corresponding instants, respectively, h is the number of data.

6. The time domain calculation method for processing the vibration signal trigger time of a steam turbine blade according to claim 5, characterized in that: The smoothing filter processing of step S4 is mean filter processing on the resampled signal data, and the method used is five-point cubic smoothing filter method, and the calculation formula is: wherein, is the filtered signal data; the number of times of the multiple smoothing filtering is 8-10 times, the filtered signal data has good smoothness, and the signal data of one blade period has one peak and one trough value.

7. The time domain calculation method for processing the vibration signal trigger time of a steam turbine blade according to claim 6, characterized in that: The specific implementation method of step S5 comprises the following steps: S5.1, the filtered signal data S5.1, the filtered signal data The calculation formula is: S5.2, then Y d Signal data is calculated for ascending and descending calibration where Y d Difference greater than 0, Y s For values of 1, Y d Difference equal to 0, Y s For values of 0, Y d Difference less than 0, Y s For values of -1; S5.3, repeat step S5.2, record Y d the time of the point whose value is equal to 2, and the Y corresponding to the time b the value, the intersection point of the upper envelope, record Y d the time of the point whose value is equal to -2, and the Y corresponding to the time b the value, the intersection point of the lower envelope, draw the intersection point of the upper envelope and the intersection point of the lower envelope on the curve, to obtain the upper envelope line and the lower envelope line.

8. The time domain calculation method for processing the vibration signal trigger time of a steam turbine blade according to claim 7, characterized in that: The coefficient of step S6 is a scaling coefficient k1, and the value of k1 is 0.7.

Images