Blade tip arrival time extraction system and method based on blade tip sensor signal centroid

Abstract

The invention discloses a blade tip arrival time extraction system and method based on the centroid of a blade tip sensor signal. A blade tip sensor is installed on a casing and is responsible for acquiring a blade tip sensor signal A; a signal conditioning module is used for transforming and adjusting the blade tip sensor signal A to obtain a blade tip sensor signal B; a data acquisition module is used for discretely quantizing the blade tip sensor signal B into a blade tip sensor signal C; a sliding mean filter module is used for performing sliding mean filtering on the blade tip sensor signal C to obtain a blade tip sensor signal D; a rotation speed estimation module provides a counting threshold; a signal cutting module dynamically cuts the blade tip sensor signal D based on the counting threshold to obtain a blade tip sensor signal E; and a centroid extraction module is used for real-time calculation of the centroid of the blade tip sensor signal E to extract the blade tip arrival time, thereby providing data support for subsequent blade vibration measurement.

Description

Technical Field

[0001] The invention relates to the field of non-contact vibration measurement of rotating blades, and in particular to a method and system for extracting blade tip arrival time based on the centroid of blade tip sensing signals. Background Art

[0002] As the core working element of large rotating machinery such as aircraft engines and gas turbines, the vibration state of moving blades affects the normal operation and working efficiency of the entire system. As a non-contact moving blade vibration measurement method, the blade tip timing method is widely used in blade vibration parameter measurement. Its measurement principle is to install a speed synchronization sensor on the stator and a blade tip sensor on the casing. When the blade tip sweeps over the blade tip sensor, a blade tip sensor signal is generated, which is converted into a blade tip arrival time pulse signal after processing. The time of the blade tip arrival time relative to the speed synchronization signal is recorded. When the blade vibrates, it will cause the blade tip arrival time to advance or lag. The blade tip arrival time pulse signal is processed by various blade tip timing processing algorithms to obtain the blade vibration parameters. Therefore, one of the key technologies for blade vibration parameter measurement based on blade tip timing is to extract the blade tip arrival time in real time with high precision, thereby providing reliable raw data for the subsequent vibration parameter identification algorithm.

[0003] However, the blade tip clearance of large rotating machinery will fluctuate within a certain range during operation, causing fluctuations in the blade tip sensor signal intensity. The complex and harsh working environment causes high-intensity random noise interference signals to be superimposed on the blade tip sensor signal, which poses a great challenge to the accuracy of blade tip arrival time extraction. The traditional blade tip arrival time extraction method only uses the edge of the blade tip sensor signal to obtain the arrival time, does not fully utilize the overall waveform information of the signal, is easily affected by signal intensity fluctuations and noise, and has low time extraction accuracy. In recent years, the full waveform processing methods developed, such as the blade tip arrival time extraction method based on ensemble empirical mode decomposition, the blade tip arrival time extraction method based on blade tip surface microstructure, and the blade tip arrival time extraction method based on phase, are all in the laboratory research stage. There are many problems such as high computational complexity and difficulty in achieving real-time processing, and it can only be applied under specific conditions. The value of industrial field application needs to be further verified. Therefore, it is urgent to design a high-precision and real-time blade tip arrival time extraction method and system to solve the above problems. Summary of the invention

[0004] The purpose of the present invention is to solve the problem of insufficient accuracy and real-time performance of the existing blade tip arrival time extraction method, and propose a blade tip arrival time extraction system and method based on the centroid of blade tip sensor signal. This method constructs a blade tip arrival time extraction algorithm model based on the centroid of blade tip sensor signal, uses the centroid of blade tip sensor signal to characterize the blade tip arrival time, makes full use of the overall waveform information of the blade tip sensor signal, and reduces the influence of signal intensity fluctuation and noise on the accuracy of blade tip arrival time extraction; constructs a blade tip arrival time extraction system structure based on the centroid of blade tip sensor signal, realizes high-precision real-time extraction of blade tip arrival time, and can be used to improve the high-precision online measurement and identification accuracy of blade vibration parameters, solves the problem of serious lack of vibration parameter data and measurement accuracy that cannot meet the requirements in the process of dynamic blade state prediction, dynamic behavior research, and failure mechanism research, provides data support for large-scale rotating machinery life prediction and dynamic blade failure mechanism research, and ensures the long-term healthy operation of equipment.

[0005] The objective of the present invention is achieved through the following technical solutions:

[0006] A blade tip arrival time extraction system based on blade tip sensor signal centroid, comprising a blade tip sensor, a signal conditioning module, a data acquisition module, a sliding mean filter module, a speed estimation module, a signal cutting module, and a centroid extraction module connected in sequence, wherein the sliding mean filter module is connected to the signal cutting module;

[0007] The blade tip sensor is installed on the casing and is responsible for obtaining the blade tip sensor signal A; the signal conditioning module is used to transform and adjust the blade tip sensor signal A to obtain the blade tip sensor signal B;

[0008] The data acquisition module is used to discretely quantify the blade tip sensor signal B into the blade tip sensor signal C;

[0009] The sliding mean filter module is used to perform sliding mean filtering on the blade tip sensor signal C to obtain the blade tip sensor signal D;

[0010] The speed estimation module estimates the speed according to the number of blades and the pulse frequency, and performs state transition to provide a counting threshold of the blade tip sensor signal D;

[0011] The signal cutting module dynamically cuts the blade tip sensor signal D based on the counting threshold to obtain the blade tip sensor signal E;

[0012] The centroid extraction module is used to calculate the centroid E of the blade tip sensor signal in real time and extract the arrival time of the blade tip, thereby providing data support for subsequent blade vibration measurement.

[0013] The present invention also provides a method for extracting the blade tip arrival time based on the blade tip arrival time extraction system, comprising:

[0014] (1) Conditioning and acquiring blade tip sensor signal B;

[0015] The blade tip sensor is installed on the casing. When the blade under test passes the blade tip sensor, a blade tip sensor signal A is generated. The signal conditioning module transforms and adjusts the blade tip sensor signal A through amplification, filtering, and noise reduction to obtain the blade tip sensor signal B.

[0016] (2) Discrete and quantize the blade tip sensor signal B into the blade tip sensor signal C through the data acquisition module;

[0017] (3) Perform sliding mean filtering on the blade tip sensor signal C to obtain the blade tip sensor signal D;

[0018] (4) Adaptively set the counting threshold;

[0019] In the speed estimation module, 10 states are set according to the speed range. Each state outputs the corresponding counting threshold, and the state is transferred according to the number of blades and pulse frequency. If the blade to be tested is not within the speed range, it is defined as a signal failure, and the subsequent modules stop working.

[0020] (5) Based on the blade tip sensor signal D obtained in step (3) and the counting threshold set in step (4), in the signal cutting module, searching for the maximum voltage value in the blade tip sensor signal D, using the counting threshold to determine the dynamic cutting position of the blade tip sensor signal D, and extracting the blade tip sensor signal E;

[0021] (6) Extract the centroid of the blade tip sensor signal E; the centroid of the blade tip sensor signal E at time t c is the time-weighted average of the amplitude of the blade tip sensor signal E, using the centroid time t of the blade tip sensor signal E c Characterizes the blade tip arrival time t b ; Further, the signal strength of the blade tip sensor signal E can be enhanced by performing an exponential operation on the blade tip sensor signal E, which can further weaken the influence of signal noise on the blade tip sensor signal E and enhance the stability of the centroid calculation of the blade tip sensor signal E. -

[0022] Furthermore, in step (1), when the blade under test passes the blade tip sensor, the blade tip clearance is d, and the blade tip sensor signal B is expressed as

[0023]

[0024]

[0025] Where g(t) represents the blade tip sensor signal B as a function of time t, ε0 is the vacuum dielectric constant, and ε r is the relative dielectric constant, K c is the signal conditioning module conversion coefficient, K sis the maximum overlapping area between the blade tip and the blade end sensor probe end surface, σ g is the scale parameter, r is the core radius, δ is the blade thickness, and v is the tip linear velocity;

[0026] The frequency spectrum distribution of g(t) is still Gaussian. Taking the frequency at 99.7%, we get the bandwidth f of the blade tip sensor signal B. BW Approximately

[0027]

[0028] From formula (3), we can see that the bandwidth of the blade tip sensor signal B is related to the rotational speed. When the rotational speed is higher, the signal edge is steeper and the signal bandwidth is higher.

[0029] Furthermore, in step (2), the bandwidth of the blade tip sensor signal B is f BW , the sampling rate of the data acquisition module is f s Must at least meet

[0030] f s >2f BW (4).

[0031] Furthermore, in step (3), the sliding mean filter takes the mean of N data points as the current data point, so the output blade tip sensor signal D is

[0032]

[0033] In the formula, s f (nt0) indicates that the blade tip sensor signal D is a sequence that changes with time nt0, where t0 is the sampling interval, n is the number of sampling points, and s i (nt0) represents the input blade tip sensor signal C over time nt0;

[0034] Cut-off frequency f c for

[0035]

[0036] The sliding mean filter module adjusts the filter cutoff frequency by selecting the number of smoothing points N to adapt to the noise reduction of the blade tip sensor signal C at different speeds; the adjacent blade tip sensor signals Ds f (nt0) difference, yes

[0037]

[0038] Equation (7) can reduce the N-1 adders used in (5) to 2 adders.

[0039] Furthermore, in step (6),

[0040] Assume that the blade tip sensor signal E is s(t). According to the definition of the centroid, the centroid of the blade tip sensor signal E at time t c It is the time-weighted average of the amplitude of the blade tip sensor signal E, and its expression is

[0041]

[0042] Where t1 is the start time of the blade tip sensor signal E, t2 is the end time of the blade tip sensor signal E, and the signal centroid time t c Characterizes the blade tip arrival time t b ;

[0043] The blade tip sensor signal E is sampled and quantized at intervals of t0, and the blade tip sensor signal E is expressed as s(nt0). The blade tip reaches the time t b for

[0044]

[0045] Where n1 is the starting point of the blade tip sensor signal E, and n2 is the ending point of the blade tip sensor signal E;

[0046] The exponential operation result of the blade tip sensor signal E is used to calculate the centroid of the blade tip sensor signal E and improve the blade tip arrival time. Expressed as

[0047]

[0048] In the formula, c represents the cth power.

[0049] Compared with the prior art, the technical solution of the present invention has the following beneficial effects:

[0050] (1) The centroid of the blade tip sensor signal is used to represent the blade tip arrival time, eliminating the influence of the signal intensity fluctuation caused by the blade tip clearance change on the accuracy of blade tip arrival time extraction;

[0051] (2) The overall waveform information of the blade tip sensor signal is utilized to reduce the influence of signal noise on the accuracy of blade tip arrival time extraction;

[0052] (3) An algorithm model and system structure for extracting the blade tip arrival time based on the centroid of the blade tip sensor signal were constructed, achieving high-precision real-time extraction of the blade tip arrival time. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 This is the structural block diagram of the high-precision real-time extraction system of the blade tip arrival time.

[0054] Figure 2 This is a structural block diagram of a sliding mean filter with dynamically adjustable cutoff frequency.

[0055] Figure 3 Simplified state diagram for the speed estimation module.

[0056] Figure 4 Schematic diagram of the centroid calculation of the blade tip sensor signal.

[0057] Figure 5 This is a simplified diagram of the centroid extraction module pipeline calculation.

[0058] Figure numerals: 1-blade tip sensor, 2-signal conditioning module, 3-data acquisition module, 4-sliding mean filter module, 5-speed estimation module, 6-signal cutting module, 7-center of mass extraction module. DETAILED DESCRIPTION

[0059] The present invention will be further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0060] This embodiment provides a blade tip arrival time extraction system based on blade tip sensor signal centroid, including a blade tip sensor 1, a signal conditioning module 2, a data acquisition module 3, a sliding mean filter module 4, a speed estimation module 5, a signal cutting module 6, and a centroid extraction module 7. The system structure block diagram is shown in FIG. Figure 1 shown.

[0061] The blade tip sensor signal generated when the blade in the blade disk to be tested sweeps over the blade tip sensor is defined as blade tip sensor signal A; the blade tip sensor signal A processed by the signal conditioning module is defined as blade tip sensor signal B; the blade tip sensor signal B collected by the signal acquisition module and discretely quantized is defined as blade tip sensor signal C; the blade tip sensor signal C filtered by the sliding mean filter module is defined as blade tip sensor signal D; the blade tip sensor signal D cut by the signal cutting module is defined as blade tip sensor signal E. Among them, blade tip sensor signals A and B are analog signals, and blade tip sensor signals C, D, and E are digital signals.

[0062] The blade tip sensor 1 is installed on the casing and is responsible for obtaining the blade tip sensor signal A; the signal conditioning module 2 is connected to the blade tip sensor 1 and is responsible for transforming, adjusting and improving the blade tip sensor signal A to improve the accuracy and sensitivity of the blade tip sensor signal A and obtain the blade tip sensor signal B; the data acquisition module 3 is connected to the signal conditioning module 2 and is responsible for discretely quantizing the blade tip sensor signal B into the blade tip sensor signal C to provide data stream for subsequent steps; the sliding mean filter module 4 is connected to the data acquisition module 3 and is responsible for sliding mean filtering the blade tip sensor signal C to further improve the signal quality of the blade tip sensor signal C. Noise ratio, obtain the blade tip sensor signal D; the speed estimation module 5 is connected to the sliding mean filter module 4, which is responsible for estimating the speed according to the number of blades and the pulse frequency, and performing state transfer to provide a counting threshold of the blade tip sensor signal D; the signal cutting module 6 is connected to the sliding mean filter module 4 and the speed estimation module 5, and is responsible for dynamically cutting the blade tip sensor signal D based on the counting threshold to extract the blade tip sensor signal E; the centroid extraction module 7 is connected to the signal cutting module 6, and is responsible for real-time calculation of the centroid of the blade tip sensor signal E, extracting the blade tip arrival time, and thus providing a data basis for subsequent blade vibration measurement.

[0063] The method for extracting the blade tip arrival time based on the centroid of the blade tip sensor signal is as follows:

[0064] (1) Conditioning and obtaining blade tip sensor signal B

[0065] The blade tip sensor 1 is installed on the casing. When the blade under test passes the blade tip sensor 1, a blade tip sensor signal A is generated. The blade tip sensor signal A is affected by noise and has burrs and large random errors. The signal conditioning module 2 transforms, adjusts and improves the blade tip sensor signal A through amplification, filtering, noise reduction, etc., and converts it into a blade tip sensor signal B.

[0066] This embodiment takes the capacitive blade tip sensor as an example. When the blade under test passes the blade tip sensor 1, the blade tip gap d is approximately constant. Ignoring the edge effect, the capacitance change formed between the blade tip sensor 1 and the blade tip is similar to a Gaussian pulse. After being processed by the signal conditioning module 2, the blade tip sensor signal B can be expressed as:

[0067]

[0068]

[0069] Where g(t) represents the blade tip sensor signal B as a function of time t, ε0 is the vacuum dielectric constant, and ε r is the relative dielectric constant, K c K is the conversion coefficient of signal conditioning module 2. s is the maximum overlapping area between the blade tip and the probe end face of blade end sensor 1, σ gis the scale parameter, r is the core radius, δ is the blade thickness, and v is the tip linear velocity.

[0070] The frequency spectrum distribution of g(t) is still Gaussian. Taking the frequency at 99.7%, we can get the bandwidth f of the blade tip sensor signal B. BW Approximately

[0071]

[0072] From formula (3), it can be obtained that the bandwidth of the blade tip sensor signal B is related to the rotation speed. When the rotation speed is higher, the signal edge is steeper and the signal bandwidth is higher. When the blade tip linear velocity v = 500m / s, the blade thickness δ = 2mm, and the core pole radius r = 2.5mm, the signal bandwidth f BW About 250kHz.

[0073] The blade tip sensor signal B obtained from the signal conditioning module 2 will be affected by the signal conditioning module 2 and the environmental interference, and a certain intensity of modulation noise and random white noise are superimposed on the blade tip sensor signal B. From formula (1), it can be obtained that when the blade tip clearance d increases, the blade tip sensor signal B will be submerged in the noise.

[0074] Therefore, in order to achieve high-precision blade tip arrival time extraction, the present invention designs a high-precision real-time extraction method for blade tip arrival time based on the centroid extraction of blade tip sensor signals. This method utilizes the full waveform information of the blade tip sensor signal to reduce the time extraction error introduced by local noise.

[0075] (2) Collecting and acquiring blade tip sensor signal C;

[0076] Furthermore, the blade tip sensor signal B needs to be discretized and quantized into the blade tip sensor signal C. From equation (3) and actual engineering applications, it can be seen that the maximum bandwidth of the blade tip sensor signal C is about 250 kHz. According to the sampling theorem, it can be obtained that the Nyquist frequency of the discrete system is required to be higher than the bandwidth f of the sampled blade tip sensor signal B. BW , signal aliasing can be avoided and the original signal can be reconstructed. The Nyquist frequency is half of the sampling frequency of the discrete system, which means that the sampling rate of the data acquisition module 3 is required to be f s At least meet

[0077] f s >2f BW (4)

[0078] But in actual engineering applications, the sampling rate f s At least the signal bandwidth f BW 7 to 10 times of that of the data acquisition module 3, so the single-channel sampling rate f s It should be greater than 2.5MSPS to meet the processing needs of subsequent steps.

[0079] (3) Sliding mean filtering to obtain the blade tip sensor signal D;

[0080] In order to provide reliable digital blade tip sensor signals for subsequent steps and reduce the interference of random noise, it is necessary to digitally filter the blade tip sensor signal C to further improve the signal-to-noise ratio. Since high-order FIR and IIR filters will occupy a large amount of DSP logic resources, and the improvement of the signal-to-noise ratio has no obvious optimization on the accuracy of the present invention, this embodiment adopts an improved form of sliding mean filter that occupies less resources.

[0081] The principle of sliding mean filtering is to take the mean of N data points as the current data point, then the output blade tip sensor signal D is

[0082]

[0083] In the formula, s f (nt0) indicates that the blade tip sensor signal D is a sequence that changes with time nt0, where t0 is the sampling interval, n is the number of sampling points, and s i (nt0) indicates that the input blade tip sensor signal C is a sequence that changes with time nt0.

[0084] The sliding mean filter has a simple structure and is easy to implement. It is suitable for smoothing burrs in waveforms. It is a classic digital low-pass filter with a cutoff frequency of f. c for

[0085]

[0086] The sliding mean filter module 4 can select the number of smoothing points N and adjust the filter cutoff frequency to adapt to the noise reduction of the blade tip sensor signal C at different speeds. The sliding mean filter algorithm is suitable for implementation on a software platform. However, in the digital logic design implementation, a large number of smoothing points will occupy a large amount of adder resources, and the filter data throughput is difficult to improve. By observing the characteristics of formula (5), the adjacent blade tip sensor signal Ds f (nt0) difference, yes

[0087]

[0088] Formula (7) can reduce the N-1 adders used in (5) to 2 adders, and can use the BRAM (Block RAM) resources in the FPGA to reduce the occupation of trigger resources. The sliding mean filter structure designed by the present invention that can dynamically adjust the cutoff frequency is as follows: Figure 2 shown.

[0089] (4) Adaptively set the counting threshold;

[0090] It can be seen from formula (3) that the frequency bandwidth of the blade tip sensor signal B is different at different speeds. The time domain shows that the pulse width of the blade tip sensor signal B changes dynamically with the speed. When the speed is higher, the pulse width is smaller, that is, the rise time of the blade tip sensor signal B is shorter. Therefore, in order to extract the blade tip sensor signal E for centroid calculation from the blade tip sensor signal D, it is necessary to set the counting threshold for judging the blade tip sensor signal D according to the speed.

[0091] like Figure 3 As shown, in the speed estimation module 5, according to the speed range, 10 states are designed, each state outputs the corresponding counting threshold, and the state transition is performed according to the number of blades and the pulse frequency. If the blade to be tested is not within the speed range, it is defined as a signal failure, and the subsequent modules stop working. Figure 3 The internal RPM represents the rotation speed, from MIN to MAX, it represents 10 states of rotation speed. When the pulse counter (pulse_cnt) is greater than the maximum value of the previous state, the state will transfer to the next state (for example, when the rotation speed is not less than the maximum value rpm_min_h in the RPM_MIN state, it will transfer to RPM_06K). When the pulse counter is less than the minimum value of the previous state, the state will transfer to the next state (for example, when the rotation speed is not greater than the minimum value rpm_06k_l in the RPM_06K state, it will transfer to RPM_MIN).

[0092] RPM is the rotation speed, and the following xx K is the pulse count value. For example, 06K means the pulse count is 6000. When the blade disk to be tested has 60 blades and the rotation speed is 100rpm, there will be 6000 pulses per minute.

[0093] (5) Extract blade tip sensor signal E;

[0094] The blade tip sensor signal D is obtained through step (3), and the counting threshold adaptively set with the rotation speed is obtained through step (4). In the signal cutting module 6, the maximum voltage value in the blade tip sensor signal D is searched, and the dynamic cutting position of the blade tip sensor signal D is determined by using the counting threshold, and the blade tip sensor signal E is extracted.

[0095] (6) Extract the centroid of the blade tip sensor signal E;

[0096] like Figure 4 As shown, let the blade tip sensor signal E be s(t). According to the definition of the centroid, the centroid of the blade tip sensor signal E is at time t c is the time-weighted average of the amplitude of the blade tip sensor signal E, and the calculation expression is:

[0097]

[0098] Where t1 is the start time of the blade tip sensor signal E, t2 is the end time of the blade tip sensor signal E, and the signal centroid time t c Characterizes the blade tip arrival time t b .

[0099] The blade tip sensor signal E is sampled and quantized at intervals of t0, and the blade tip sensor signal E can be expressed as s(nt0). The blade tip arrival time t b for

[0100]

[0101] Where n1 is the starting time point of the blade tip sensor signal E, and n2 is the ending time point of the blade tip sensor signal E.

[0102] Furthermore, in order to further improve the extraction accuracy of the blade tip arrival time of the present invention, the technical solution can be improved by increasing the strength of the blade tip sensor signal E, that is, using the exponential operation result of the blade tip sensor signal E to calculate the centroid of the blade tip sensor signal E, which can further weaken the influence of signal noise on the blade tip sensor signal E and enhance the stability of the calculation of the centroid of the blade tip sensor signal E. The improved version of the blade tip arrival time It can be expressed as

[0103]

[0104] In the formula, c represents the cth power.

[0105] In order to realize the real-time extraction of the blade tip arrival time, the present invention adopts FPGA using pipeline optimization technology. According to formula (10), the calculation of the blade tip arrival time can be divided into four pipeline stages, a non-pipeline division stage, and a time pulse generation stage. Figure 5 The figure shows the logic block diagram of FPGA. The upper part of the figure is an introduction to each stage, and the lower part is the specific operation process of FPGA. The four pipeline stages are to calculate the absolute value, square, weighted value of each point of the blade tip sensor signal E, and the sum of each point of the blade tip sensor signal E. A non-pipeline division stage is to calculate the centroid of the blade tip sensor signal E. Finally, a time pulse is generated based on the centroid, providing a real-time and high-precision data basis for subsequent blade vibration measurement.

[0106] Finally, it should be pointed out that the above examples are only used to illustrate the calculation process of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the above examples, a person skilled in the art should understand that the calculation process described in the above examples can still be modified, or some parameters can be replaced by equivalents, and these modifications or replacements do not make the essence of the corresponding calculation method deviate from the spirit and scope of the calculation method of the present invention.

[0107] The present invention is not limited to the embodiments described above. The above description of the specific embodiments is intended to describe and illustrate the technical solution of the present invention. The above specific embodiments are merely illustrative and not restrictive. Without departing from the scope of the present invention and the scope of protection of the claims, a person of ordinary skill in the art can also make many forms of specific changes under the guidance of the present invention, which all fall within the scope of protection of the present invention.

Claims

1. A blade tip arrival time extraction system based on the centroid of blade tip sensor signals, characterized in that: It comprises a blade tip sensor (1), a signal conditioning module (2), a data acquisition module (3), a sliding mean filter module (4), a rotation speed estimation module (5), a signal cutting module (6), and a centroid extraction module (7) which are connected in sequence, wherein the sliding mean filter module (4) is connected to the signal cutting module (6); The blade tip sensor (1) is installed on the casing and is responsible for obtaining a blade tip sensor signal A; the signal conditioning module (2) is used to transform and adjust the blade tip sensor signal A to obtain a blade tip sensor signal B; The data acquisition module (3) is used to discretely quantize the blade tip sensor signal B into the blade tip sensor signal C; The sliding mean filter module (4) is used to perform sliding mean filtering on the blade tip sensor signal C to obtain the blade tip sensor signal D; The rotation speed estimation module (5) estimates the rotation speed according to the number of blades and the pulse frequency, performs state transfer, and provides a counting threshold of the blade tip sensor signal D; The signal cutting module (6) dynamically cuts the blade tip sensor signal D based on the counting threshold to obtain the blade tip sensor signal E; The centroid extraction module (7) is used to calculate the centroid of the blade tip sensor signal E in real time, extract the blade tip arrival time, and further provide data support for subsequent blade vibration measurement.

2. A method for extracting blade tip arrival time based on the blade tip arrival time extraction system according to claim 1, characterized in that: include: (1) Conditioning and acquiring blade tip sensor signal B; The blade tip sensor is installed on the casing. When the blade under test passes the blade tip sensor, a blade tip sensor signal A is generated. The signal conditioning module transforms and adjusts the blade tip sensor signal A through amplification, filtering, and noise reduction to obtain the blade tip sensor signal B. (2) Discrete and quantize the blade tip sensor signal B into the blade tip sensor signal C through the data acquisition module; (3) Perform sliding mean filtering on the blade tip sensor signal C to obtain the blade tip sensor signal D; (4) Adaptively set the counting threshold; In the speed estimation module, 10 states are set according to the speed range. Each state outputs the corresponding counting threshold, and the state is transferred according to the number of blades and pulse frequency. If the blade to be tested is not within the speed range, it is defined as a signal failure, and the subsequent modules stop working. (5) Based on the blade tip sensor signal D obtained in step (3) and the counting threshold set in step (4), in the signal cutting module, searching for the maximum voltage value in the blade tip sensor signal D, using the counting threshold to determine the dynamic cutting position of the blade tip sensor signal D, and extracting the blade tip sensor signal E; (6) Extract the centroid of the blade tip sensor signal E; the centroid of the blade tip sensor signal E at time t c is the time-weighted average of the amplitude of the blade tip sensor signal E, using the centroid time t of the blade tip sensor signal E c Characterizes the blade tip arrival time t b .

3. According to claim 2, a method for extracting blade tip arrival time based on the blade tip arrival time extraction system according to claim 1, characterized in that: In step (1), when the blade under test passes the blade tip sensor, the blade tip gap is d, and the blade tip sensor signal B is expressed as Where g(t) represents the blade tip sensor signal B as a function of time t, ε0 is the vacuum dielectric constant, and ε r is the relative dielectric constant, K c is the signal conditioning module conversion coefficient, K s is the maximum overlapping area between the blade tip and the blade end sensor probe end surface, σ g is the scale parameter, r is the core radius, δ is the blade thickness, and v is the tip linear velocity; The frequency spectrum distribution of g(t) is still Gaussian. Taking the frequency at 99.7%, we get the bandwidth f of the blade tip sensor signal B. BW Approximately From formula (3), we can see that the bandwidth of the blade tip sensor signal B is related to the rotational speed. When the rotational speed is higher, the signal edge is steeper and the signal bandwidth is higher.

4. According to claim 2, a method for extracting blade tip arrival time based on the blade tip arrival time extraction system according to claim 1, characterized in that: In step (2), the blade tip sensor signal B bandwidth is f BW , the sampling rate of the data acquisition module is f s Must at least meet in s >2f BW (4).

5. According to claim 2, a method for extracting blade tip arrival time based on the blade tip arrival time extraction system according to claim 1, characterized in that: In step (3), the sliding mean filter takes the mean of N data points as the current data point, so the output blade tip sensor signal D is In the formula, s f (nt0) indicates that the blade tip sensor signal D is a sequence that changes with time nt0, where t0 is the sampling interval, n is the number of sampling points, and s i (nt0) indicates that the input blade tip sensor signal C is a sequence that changes with time nt0; Cut-off frequency f c for The sliding mean filter module adjusts the filter cutoff frequency by selecting the number of smoothing points N to adapt to the noise reduction of the blade tip sensor signal C at different speeds; the adjacent blade tip sensor signals Ds f (nt0) difference, yes Formula (7) can reduce the N-1 adders used in (5) to 2 adders.

6. According to claim 2, a method for extracting blade tip arrival time based on the blade tip arrival time extraction system according to claim 1, characterized in that: In step (6), Assume that the blade tip sensor signal E is s(t). According to the definition of the centroid, the centroid of the blade tip sensor signal E at time t c It is the time-weighted average of the amplitude of the blade tip sensor signal E, and its expression is Where t1 is the starting time of the blade tip sensor signal E, t2 is the ending time of the blade tip sensor signal E, and the centroid time t c Characterizes the blade tip arrival time t b ; The blade tip sensor signal E is sampled and quantized at intervals of t0, and the blade tip sensor signal E is expressed as s(nt0). The blade tip reaches the time t b for Where n1 is the starting point of the blade tip sensor signal E, and n2 is the ending point of the blade tip sensor signal E; The exponential operation result of the blade tip sensor signal E is used to calculate the centroid of the blade tip sensor signal E and improve the blade tip arrival time. Expressed as In the formula, c represents the cth power.

7. A method for extracting blade tip arrival time based on the blade tip arrival time extraction system according to claim 2, characterized in that: In step (6), the signal strength of the blade tip sensor signal E is enhanced by exponentially operating the blade tip sensor signal E to further weaken the influence of signal noise on the blade tip sensor signal E and enhance the stability of the centroid calculation of the blade tip sensor signal E.

Images