A method for diagnosing a gas turbine blade failure based on a sound pressure signal

Abstract

The present application aims at providing a gas turbine blade fault diagnosis method based on sound pressure signal, belonging to the field of gas turbine diagnosis, comprising the following steps: monitoring the sound pressure of the gas turbine, collecting the rotating speed of the gas turbine; sequentially storing the rotating speed of the gas turbine; giving a gas turbine steady state criterion, recording the noise monitoring value when the gas turbine is in steady state operation condition; calculating the frequency spectrum of the sound pressure of the gas turbine; calculating the rotating speed and rotating frequency according to the key belief signal; calculating the passing frequency of the blade under a specific rotating speed, and obtaining the amplitude corresponding to the passing frequency of each stage of blade; according to different rotating speeds, taking the boundary band value ± epsilon as the alarm threshold; segmenting and modeling the alarm threshold under different rotating speeds, constructing the sound pressure alarm limit band of the gas turbine under different operating rotating speeds; when the characteristic frequency exceeds the alarm limit, the corresponding alarm is given. The present application solves the problems of equipment operation state monitoring under the condition that the vibration sensor is not added to the current gas turbine.

Description

Technical Field

[0001] This invention relates to a gas turbine diagnostic method, specifically a blade diagnostic method. Background Technology

[0002] Gas turbines, as an important type of power machinery, are characterized by their compact structure, stable operation, and high thermal efficiency, leading to their increasingly wide range of applications. The safe and reliable operation of gas turbines requires high standards. Monitoring the health of the unit during daily operation and analyzing and detecting potential anomalies can help prevent or facilitate the timely handling of major gas turbine failures.

[0003] A gas turbine consists of two main components: the compressor and the turbine, both composed of multiple stages of blades. During operation, the high-speed airflow passing through the blades generates noise. The noise frequency produced by the airflow passing through each stage of the gas turbine blade cascade varies depending on the number of blades, and the noise also differs at different airflow speeds. To effectively extract the characteristics of the airflow passing through each stage of the blade cascade from these noises and to achieve gas turbine blade fault diagnosis, a universal method for extracting the passing frequencies of gas turbine blades is needed. Summary of the Invention

[0004] The purpose of this invention is to provide a gas turbine blade fault diagnosis method based on sound pressure signals that can solve the problems of equipment operation status monitoring under the current restrictions on installing vibration sensors in gas turbines.

[0005] The objective of this invention is achieved as follows:

[0006] This invention discloses a method for diagnosing gas turbine blade faults based on sound pressure signals, characterized by comprising the following steps:

[0007] (1) Install a sound pressure sensor on the compressor side of the gas turbine to realize the sound pressure monitoring of the gas turbine; read the key phase signal of the gas turbine speed to realize the acquisition of the gas turbine speed;

[0008] (2) Store the gas turbine speed sequentially and use the first-in-first-out principle to obtain the gas turbine speed sequence;

[0009] (3) Given the steady-state criterion for gas turbines, As a steady-state criterion, where n max n is the maximum value of the rotational speed sequence. min The minimum value of the speed sequence is given; when e < 1%, the gas turbine is in steady-state operation, and the gas turbine noise monitoring value is recorded at this time.

[0010] (4) The spectrum M of the gas turbine sound pressure is calculated using the Fast Fourier Transform. i ;

[0011] (5) According to the key signal, the rotation speed n and the rotation frequency f are calculated r , assuming that the speed measuring gear is m teeth, then:

[0012] fr=n / m;

[0013] (6) Selecting an arbitrary first-stage blade, the through frequency of the blade at a specific rotation speed is calculated according to the number of blades, and the amplitude corresponding to the through frequency of each stage blade is obtained;

[0014] (7) According to the historical operation data of the amplitude corresponding to the through frequency of each stage blade collected in the normal operation state, statistics are made according to different rotation speeds, and the boundary band value ± ε is taken as the alarm threshold;

[0015] (8) The alarm threshold at different rotation speeds is segmented and modeled in steps, and the sound pressure alarm limit band of the gas turbine at different operation rotation speeds is constructed;

[0016] (9) When the characteristic frequency exceeds the alarm limit, the corresponding alarm is given, and the gas turbine blade fault diagnosis based on the sound pressure signal is realized.

[0017] The application can also include:

[0018] 1. The maximum value and the minimum value of the rotation speed sequence in step (3) are found by the linear scanning method:

[0019] A. Initialize two variables: value_max and value_min, and give the amplitude of the first element of the rotation speed sequence to the two variables;

[0020] B. Traverse the rotation speed array, compare the current iteration value with value_max and value_min each time, if it is greater than value_max and less than value_min, update the variable value, if not, continue iteration;

[0021] C. After the traversal is completed, the values of the two variables are the maximum value and the minimum value of the rotation speed sequence.

[0022] 2. After obtaining the rotation frequency in step (5), search for the peak value in the range of fr±fr*0.1 near the rotation frequency, and obtain the first rotation frequency f r ; calculate each frequency domain feature in the following interval through the first rotation frequency:

[0023] The amplitude of the rotation frequency: P1=max(M fr-f r *0.1 , M fr+f r *0.1 )

[0024] The amplitude of the second rotation frequency: P2=max(M 2fr-fr*0.1 , M 2f r +f r*0.1 )

[0025] The amplitude of the three times frequency conversion: P3 = max (M 3fr-fr*0.1 , M 3f r +f r *0.1 )

[0026] The amplitude of the four times frequency conversion: P4 = max (M 4fr-fr*0.1 , M 4f r +f r *0.1 )

[0027] The amplitude of the one half frequency division: P5 = max (M 1 / 2fr-fr*0.05 , M 1 / 2f r +f r *0.05 )

[0028] The amplitude of the one third frequency division: P6 = max (M 1 / 3fr-fr*0.05 , M 1 / 3f r +f r *0.05 ).

[0029] 3, the amplitude search method is the insertion sort method, the specific steps are as follows:

[0030] ① from the first element, the element is considered to have been sorted;

[0031] ② take out the next element, scan from back to front in the sorted element sequence;

[0032] ③ if the element is greater than the new element, move the element to the next position;

[0033] ④ repeat step ③ until the sorted element less than or equal to the new element is found;

[0034] ⑤ after the new element is inserted into the position;

[0035] ⑥ repeat steps ③-⑤ until all elements are sorted;

[0036] ⑦ according to the sorted array, the maximum amplitude in each frequency range is obtained.

[0037] 4, the selection method of the amplitude corresponding to the frequency of each level of blade in step (6) is:

[0038]

[0039] 5, step (8) is specifically:

[0040] i, take 50 RPM as an interval, obtain the alarm threshold of 50 to 5000 RPM speed;

[0041] ii. According to the obtained alarm threshold point, an alarm threshold polyline is drawn, and a sound pressure alarm limit band is obtained.

[0042] The present application has the advantage that it solves the problem of monitoring the running state of equipment under the condition that the vibration sensor is limited to be added to the gas turbine. BRIEF DESCRIPTION OF DRAWINGS

[0043] Figure 1 An illustration for data device is provided;

[0044] Figure 2 An illustration for step (6) spectrum extraction is provided;

[0045] Figure 3 An illustration for step (8) sound pressure alarm limit band is provided;

[0046] Figure 4 A flow chart of the present application is provided. DETAILED DESCRIPTION

[0047] The present application will be described in more detail below with examples and drawings:

[0048] In combination Figures 1-4 , the steps of the present application are as follows:

[0049] Step 1: According to the specific situation of the gas turbine, a sound pressure sensor is arranged at a distance of about 1 meter on the side of the gas turbine compressor to realize sound pressure monitoring of the gas turbine; a gas turbine speed key signal is read to realize collection of the gas turbine speed.

[0050] Step 2: The gas turbine speed is sequentially stored, and the first-in-first-out principle is adopted to obtain a gas turbine speed sequence.

[0051] Step 3: A steady state criterion of the gas turbine is given. as the steady state criterion, wherein n max is the maximum value of the speed sequence, and n min is the minimum value of the speed sequence. When e < 1%, it is considered that the gas turbine is in a steady state running condition, and the noise monitoring value of the gas turbine is recorded at this time.

[0052] The maximum and minimum values of the speed sequence are found by linear scanning method, and the specific steps are as follows:

[0053] 1. Initialize two variables, value_max and value_min, and give the first element amplitude of the speed sequence to the two variables;

[0054] 2. Traverse the array of rotational speed, compare the current iteration value with value_max and value_min each time, if yes, update the variable value, if not, continue iteration;

[0055] 3. After the completion of the traversal, the values of the two variables are the maximum and minimum values of the rotational speed sequence.

[0056] Step 4: Use the fast Fourier transform to calculate the frequency spectrum M of the sound pressure of the gas turbine i .

[0057] Step 5: Calculate the rotational speed n and the rotational frequency fr according to the key phase signal. Assuming that the speed measuring toothed disc has m teeth, then:

[0058] f r =n / m

[0059] After obtaining the rotational frequency, perform peak value search in the range of f r ±f r *0.1 near the rotational frequency to obtain the first-order rotational frequency f r . Calculate each frequency domain feature in the following interval through the first-order rotational frequency.

[0060] The amplitude of the rotational frequency: P1 = max(M fr-fr*0.1 , M fr+f r *0.1 )

[0061] The amplitude of the second-order rotational frequency: P2 = max(M 2fr-fr*0.1 , M 2f r +f r *0.1 )

[0062] The amplitude of the third-order rotational frequency: P3 = max(M 3fr-fr*0.1 , M 3f r +f r *0.1 )

[0063] The amplitude of the fourth-order rotational frequency: P4 = max(M 4fr-fr*0.1 , M 4f r +f r *0.1 )

[0064] The amplitude of the one-half sub-frequency: P5 = max(M 1 / 2fr-fr*0.05 , M 1 / 2f r +f r *0.05 )

[0065] The amplitude of the one-third sub-frequency: P6 = max(M 1 / 3fr-fr*0.05 , M 1 / 3f r +fr*0.05 )

[0066] Wherein the amplitude search method is the insertion sort method, and the specific steps are as follows:

[0067] 1. Start from the first element, which can be considered to have been sorted;

[0068] 2. Take the next element, and scan from the back to the front in the sorted element sequence;

[0069] 3. If the element (sorted) is greater than the new element, move the element to the next position;

[0070] 4. Repeat step 3 until the sorted element less than or equal to the new element is found;

[0071] 5. Insert the new element after the position;

[0072] 6. Repeat steps 2-5 until all elements are sorted;

[0073] 7. According to the sorted array, the maximum amplitude in each frequency range is obtained.

[0074] Step 6: Select any level blade, calculate the passing frequency of the blade at a certain speed according to the number of blades, and obtain the amplitude corresponding to the passing frequency of each level blade.

[0075] The selection method of the amplitude corresponding to the passing frequency of each level blade is:

[0076]

[0077] Step 7: According to the historical operation data of the amplitude corresponding to the passing frequency of each level blade collected under normal operation state, statistics are made according to different speeds, and (boundary band value ± ε) is taken as the alarm threshold (wherein ε is 1-5 times the standard deviation of the statistical result according to experience).

[0078] Step 8: Segment the alarm threshold at different speeds to build a ladder model, and construct a sound pressure alarm limit band of the gas turbine at different operating speeds. The specific steps are as follows:

[0079] 1. Take 50 RPM as an interval to obtain the alarm threshold at 50 to 5000 RPM;

[0080] 2. According to the obtained alarm threshold points, draw an alarm threshold line graph to obtain the sound pressure alarm limit band.

[0081] Step 9: When one or more characteristic frequencies exceed the alarm limit, give the corresponding alarm to realize the gas turbine blade fault diagnosis based on the sound pressure signal.

Claims

1. A method for diagnosing a fault of a gas turbine blade based on a sound pressure signal, characterized by: It comprises the following steps: (1) arranging an acoustic pressure sensor on the gas turbine compressor side to realize gas turbine acoustic pressure monitoring; reading a gas turbine speed key signal to realize gas turbine speed acquisition; (2) sequentially storing the gas turbine speed, and obtaining a gas turbine speed sequence by using the first-in first-out principle; (3) Given gas turbine steady state criterion, to as steady state criterion, where n max is the maximum value of the speed sequence, n min is the minimum value of the speed sequence; When e < 1%, the gas turbine is in a steady state operating condition, and the gas turbine noise monitoring value is recorded at this time; (4) Using fast Fourier transform, calculate the frequency spectrum M of the sound pressure of the gas turbine i ; (5) The rotation speed n and the rotation frequency f are calculated according to the key phase signal r If the speed measuring gear wheel has m teeth, then: f r = n / m; (6) selecting an arbitrary stage blade, calculating the passing frequency of the blade at a specific speed according to the number of blades, and obtaining the amplitude corresponding to the passing frequency of each stage blade; The selection method of the amplitude corresponding to the passing frequency of each stage blade in step (6) is: (7) according to the historical operation data of the amplitude corresponding to the passing frequency of each stage blade collected in the normal operation state, the data is counted according to different speeds, and the boundary band value ± ε is taken as the alarm threshold; (8) segmenting and modeling the alarm threshold at different speeds to construct an acoustic pressure alarm limit band of the gas turbine at different operating speeds; (9) when the characteristic frequency exceeds the alarm limit, the corresponding alarm is given to realize the gas turbine blade fault diagnosis based on the acoustic pressure signal.

2. A method of diagnosing a fault in a gas turbine blade based on a sound pressure signal according to claim 1, characterized in that: The maximum value and the minimum value of the speed sequence in step (3) are found by linear scanning method: A, initialize two variables: value_max, value_min, and give the amplitude of the first element of the speed sequence to the two variables; B, traverse the speed array, compare the current iteration value with value_max and value_min each time, if it is greater than value_max and less than value_min, update the variable value, if not, continue iteration; C, after the traversal is completed, the values of the two variables are the maximum and minimum values of the speed sequence.

3. A method of diagnosing a fault in a gas turbine blade based on a sound pressure signal according to claim 1, characterized in that: After the conversion frequency is obtained in step (5), a peak search is performed in the range of f r ±f r 0.1 around the conversion frequency to obtain a one-fold conversion frequency f r Each frequency domain feature is calculated in the following interval respectively through the one-fold conversion frequency: Amplitude of the frequency conversion: P1 = max(M fr-fr*0.1 , M fr+fr*0.1 ) Amplitude of the double frequency: P2 = max(M 2fr-fr*0.1 , M 2fr+fr*0.1 ) Amplitude of the third harmonic: P3 = max(M 3fr-fr*0.1 , M 3fr+fr*0.1 ) Amplitude of the four times frequency: P4 = max(M 4fr-fr*0.1 , M 4fr+fr*0.1 ) Half frequency divider amplitude: P5 = max(M 1 / 2fr-fr*0.05 , M 1 / 2fr+fr*0.05 ) Amplitude of the third fractional division: P6 = max(M 1 / 3fr-fr*0.05 , M 1 / 3fr+fr*0.05 ).

4. A method of diagnosing a fault in a gas turbine blade based on a sound pressure signal according to claim 3, characterized in that: The amplitude search method is the insertion sort method, and the specific steps are as follows: ① Start from the first element, which is considered to have been sorted; ② Take out the next element and scan from back to front in the already sorted element sequence; ③ If the element is greater than the new element, move the element to the next position; ④ repeat step ③ until the sorted element is less than or equal to the new element; ⑤ insert the new element after the position; ⑥ repeat steps ③-⑤ until all elements are sorted; ⑦ according to the sorted array, obtain the maximum amplitude in each frequency range.

5. A method of diagnosing a fault in a gas turbine blade based on a sound pressure signal according to claim 1, characterized in that: Step (8) is specifically: i. Obtain the 50 to 5000 RPM speed alarm threshold at intervals of 50 RPM; ii. Draw an alarm threshold line graph according to the obtained alarm threshold points to obtain the acoustic pressure alarm limit band.

Images