Water turbine vibration monitoring sensor system and method

Abstract

The invention provides a water turbine vibration monitoring sensor system and method. The water turbine vibration monitoring sensor system comprises a space compensation mechanism, a stabilizing mechanism, a distance measuring sensor and a reflection target. The reflection target is fixedly arranged on a tested device and is used for transmitting vibration; the space compensation mechanism is connected with a tested device, the stabilizing mechanism is arranged on the space compensation mechanism, and the space compensation mechanism is used for compensating the influence of vibration on the stabilizing mechanism; the stabilizing mechanism is used for keeping the relative stability of the spatial position, the distance measuring sensor is arranged in the stabilizing mechanism, and the distance measuring sensor detects the distance data of the spatial component between the distance measuring sensor and the reflection target so as to obtain the low-frequency vibration data of the detected device. According to the invention, an active driving space compensation mechanism is arranged and is matched with a prediction compensation scheme based on artificial intelligence, so that the problem of sensor stability of a conventional stabilizing mechanism in a low-frequency state is solved. And the laser range finders are adopted to respectively collect the distance between the laser range finders and the reflection target, and displacement data are accurately obtained, so that accurate low-frequency vibration data are obtained.

Description

Technical Field

[0001] The present invention relates to the technical field of turbine vibration monitoring, and particularly relates to a turbine vibration monitoring sensor system and method. Background Art

[0002] The measurement of absolute vibration, especially the measurement of ultra-low frequency absolute vibration with a vibration frequency of 1 Hz or less, remains one of the difficult problems in the field of engineering measurement to date. This is because absolute vibration is a vibration relative to inertial space without a stationary reference point. It is difficult to find a stationary reference benchmark during measurement. Also, due to the low vibration frequency, it is difficult to make the mechanical natural frequency of existing vibration sensors low enough. As a result, when measuring ultra-low frequency absolute vibration, the signal-to-noise ratio of the sensor is low, and its signal output is extremely weak, completely "submerged" in the noise and difficult to pick up. In addition, since the low-frequency vibration acceleration value is not large, it is difficult to collect data using acceleration sensors. For hydro-generator units, the vibration of the unit is an important parameter reflecting the operating state of the unit. It depends on the design, manufacturing level, and installation quality of the unit, and affects the stable and reliable operation of the unit. As the single-unit capacity of the unit increases, the structural size of the unit increases correspondingly, and the relative mechanical stiffness of the unit decreases, making the vibration problem more prominent. Through research, the factors causing the vibration of hydro-generator units include: 1. Unbalance of the mass of rotating components. When there is uneven mass distribution in rotating components such as the rotor or runner, it causes the center of gravity of the rotating component to deviate from the rotation centerline, resulting in an eccentricity of the center of gravity with respect to the centerline. A centrifugal force is generated during rotation, leading to imbalance (Ren Jishun et al., 2018). Taking the main shaft as an example, the main shaft undergoes bending deformation and generates "bow-shaped gyration". Its amplitude is proportional to the square of the rotational frequency of the unit. The vibration frequency usually equals the rotational frequency or a multiple of the rotational frequency, and the vibration direction is roughly the horizontal radial direction. The axis locus shows an ellipse with a small eccentricity. 2. Misalignment of the main shaft axis. Misalignment of the main shaft axis of the unit usually occurs at the coupling part. The root cause is the existence of a translational offset or angular deviation between the generator shaft and the runner main shaft connected by the coupling, resulting in radial or axial vibration when the unit main shaft rotates (Guo Chunhua and Peng Tao, 2009). 3. Rubbing between the moving and stationary components of the rotor or runner. When there is non-concentricity or defects in the rotating and stationary components of the unit, it will cause rubbing faults between the moving and stationary components due to mechanical factors. In addition, when the water flow out of the runner blades is asymmetric, it will also cause rubbing between the rotating component and the supporting component, accompanied by a violent impact sound (Zhang Jinjian et al., 2021). Its vibration frequency usually includes the rotational frequency and multiples of the rotational frequency, and also includes low multiple harmonic components such as (0~0.99)∙fR. The axis locus presents multiple serrated sharp corners or small loops. This type of vibration is usually a concurrent vibration of misalignment and imbalance. 4. Other mechanical factors. The vibration of the unit caused by mechanical factors also includes defects in guide bearings, excessive bearing clearance or too thin main shaft, lack of accuracy of thrust bearings, and poor fastening of the main shaft flange during the installation process of the unit. 5. Hydrodynamic factors: including uneven flow velocity distribution in the flow passage, formation of eddies, skew and swirl of the inlet water flow, etc. Especially during partial load operation, a rotating stall phenomenon may occur, such as "surge", which will cause significant vibration. 6. Factors of changes in operating conditions. The operating parameters of the water turbine under different operating conditions, such as flow rate, head, etc., will also affect its stability.Rapid load changes or frequent starts and stops can cause impacts on equipment, which in turn lead to vibrations. 7. Draft tube pressure fluctuation factor: In Francis or Kaplan turbines, the pressure pulsation inside the draft tube is caused by the combined action of the runner blade passing frequency and the guide vane opening. In severe cases, it can cause strong vibrations of the entire unit. The vibrations caused by these factors are usually obvious at low rotational speeds. Therefore, the vibration frequency range of the hydroturbine unit in actual operation is wide and the directionality is uncertain. Especially, the vibration frequency caused by hydraulic factors due to water flow vortices is approximately between 1 / 6 and 1 / 3 of the rotational frequency, mostly below 0.5 Hz. Moreover, the on-site test environment is complex, with electromagnetic interference, temperature, humidity and other factors interfering. Therefore, in order to grasp the true vibration situation of the equipment, accurately analyze the operating conditions of the unit, and then reverse-deduce the causes of vibrations to avoid equipment damage and improve work efficiency, sensors with reliable performance and accurate data are required. Chinese patent document CN102901556A records a magnetically levitated ultra-low frequency vibration sensor, which is provided with a cylindrical permanent magnet, a circular permanent magnet, an electromagnetic coil and a photoelectric displacement detector, and obtains low-frequency vibration data by detecting the relative displacement between the photoelectric displacement detector and the levitated circular permanent magnet. However, during the actual operation process, affected by vibrations, the magnetic field will also lose balance due to the eccentricity between the cylindrical permanent magnet and the circular permanent magnet, resulting in inaccurate detection data. Further, this document also proposes to adjust the electromagnetic force by using a PID control strategy to balance the system. However, due to the non-linear characteristics of the electromagnetic force, accurate input parameters cannot generate accurate levitation force, and this adjustment leads to the generation of new unstable factors, affecting the final detection accuracy. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to provide a vibration monitoring sensor system for a hydroturbine, which can accurately collect and monitor the vibration data of the hydroturbine, and provide a solid foundation for subsequent working condition analysis, operation and maintenance planning and fault prevention.

[0004] The technical problem to be solved by the present invention is to provide a monitoring method for a vibration monitoring sensor system of a hydroturbine, to further improve the accuracy of vibration data collection, and to propose a compensation control method, which can reduce the influence of vibrations on the data collection accuracy.

[0005] To solve the above technical problems, the technical solution adopted by the present invention is: A vibration monitoring sensor system for a hydroturbine, comprising a space compensation mechanism, a stabilization mechanism, a ranging sensor and a reflection target; The reflection target is fixedly arranged on the device to be measured and is used for transmitting vibrations; The space compensation mechanism is connected to the device to be measured, the stabilization mechanism is arranged on the space compensation mechanism, and the space compensation mechanism is used for compensating the influence of vibrations on the stabilization mechanism; The stabilization mechanism is used to maintain the relative stability of the spatial position. The ranging sensor is arranged inside the stabilization mechanism. The ranging sensor detects the distance data of the spatial component between it and the reflection target to obtain the low-frequency vibration data of the device under test.

[0006] In a preferred solution, the spatial compensation mechanism includes an x-direction compensation stage, a y-direction compensation stage, and a z-direction compensation stage that are stacked on top of each other; Compensation elements are provided between the x-direction compensation stage, the y-direction compensation stage, and the z-direction compensation stage, and the relative movement between the compensation stages is driven by the compensation elements.

[0007] In a preferred solution, in the spatial compensation mechanism, the base is fixedly connected to the device under test. The x-direction compensation stage is horizontally slidably arranged on the base. An x-direction hook extending towards the base is provided along the x-direction on the x-direction compensation stage. An x-direction compensation element is provided between the x-direction hook and the base; The y-direction compensation stage is horizontally slidably arranged on the x-direction compensation stage. A y-direction hook extending towards the x-direction compensation stage is provided along the y-direction on the y-direction compensation stage. A y-direction compensation element is provided between the y-direction hook and the x-direction compensation stage; The z-direction compensation stage is arranged on the y-direction compensation stage. A z-direction compensation element is provided between the z-direction compensation stage and the y-direction compensation stage; The compensation element includes a piezoelectric element or a linear motor.

[0008] In a preferred solution, the compensation element is a multi-stack piezoelectric element, and each piezoelectric element is independently driven by a driving device.

[0009] In a preferred solution, the stabilization mechanism includes an inner frame and an outer frame; The outer frame is horizontally slidably arranged on the top of the z-direction compensation stage. A plurality of horizontally arranged horizontal elastic elements are provided along the circumference between the z-direction compensation stage and the outer frame. The plurality of horizontal elastic elements are balanced and align the outer frame and the z-direction compensation stage in the center; The inner frame is slidably connected to the outer frame in the vertical direction. A plurality of vertical elastic elements are provided along the circumference between the inner frame and the outer frame. The vertical elastic elements are balanced with the self-weight of the inner frame; At least three laser rangefinders respectively point towards the reflection target along the x-direction, y-direction, and z-direction. The reflection target is provided with corresponding vertical reflection surfaces; A triaxial acceleration sensor is provided on the z-direction compensation stage.

[0010] In a preferred solution, the outer frame is a cavity structure. An opening is provided at the center of the outer frame. A plurality of barbs are provided along the circumference on the z-direction compensation stage. The barbs pass through the opening. A horizontal elastic element is provided between the barbs and the outer frame; Vertical guide posts are provided along the circumference on the inner wall of the outer frame. Corresponding sliding sleeves are provided on the inner frame. The sliding sleeves are slidably connected to the vertical guide posts; A plurality of cantilever openings are provided along the circumference on the side wall of the inner frame. Corresponding upper cantilevers are provided along the circumference on the outer frame. The upper cantilevers pass through the cantilever openings and extend into the inner frame. Lower cantilevers are provided along the circumference in the inner frame. A vertical elastic element is arranged between the upper cantilevers and the lower cantilevers.

[0011] In a preferred embodiment, the reflection target is fixedly connected to the device under test through a conduction seat. The x-direction laser rangefinder and the y-direction laser rangefinder are horizontally arranged. The x-direction laser rangefinder and the y-direction laser rangefinder are 90° apart on the circumference. The x-direction laser rangefinder and the y-direction laser rangefinder respectively vertically point to two mutually perpendicular vertical planes of the reflection target. The z-direction laser rangefinder is arranged at the bottom of the sensor top cover of the inner frame, and the z-direction laser rangefinder vertically points to the horizontal top surface of the reflection target.

[0012] In a preferred embodiment, a magnetic levitation mechanism is provided between the bottom of the outer frame and the top of the z-direction compensation table. The magnetic levitation mechanism is an array of magnets with opposite polarities. In a preferred embodiment, a counterweight is provided in the inner frame so that the center of mass of the inner frame coincides with the central axis of the inner frame.

[0013] In a preferred embodiment, a vibration sensor is provided in the reflection target to collect vibration data of the device under test.

[0014] In a preferred embodiment, a main control device and a host computer are further provided. The triaxial acceleration sensor is electrically connected to the main control device to collect the spatial displacement of the z-direction compensation table. The main control device is electrically connected to the compensation element to compensate for the influence of vibration on the spatial position of the z-direction compensation table. The laser rangefinder sends the collected data to the host computer through the main control device for analyzing vibration data.

[0015] A monitoring method using the above water turbine vibration monitoring sensor system includes the following steps: S1. In the shutdown state of the device under test, set the ranging data between the laser rangefinder and the reflection target to zero. S2. Start the device under test, and the laser rangefinder collects the distance data between it and the reflection target. S3. A triaxial acceleration sensor is provided on the z-direction compensation table to obtain the acceleration data of the inner frame and convert it into real-time displacement data. S4. A main control device is further provided. The main control device decomposes the real-time displacement data into x-direction, y-direction, and z-direction displacement data, and respectively and real-time controls the x-direction compensation element, y-direction compensation element, and z-direction compensation element to move with opposite displacement amounts to compensate for the influence of vibration on the spatial position of the inner frame. S5. The main control device stores the ranging data and time data of the laser rangefinder. The monitoring of the low-frequency vibration of the water turbine is achieved through the above steps.

[0016] Preferably, in step S4, the following steps are included: S41. Predict future vibration data using a long short-term memory network; S42. Evaluate the delay time data between data acquisition and compensation execution of the system using the phase spectrum method or the autoregressive model method; S43. Perform compensation according to the predicted future vibration data based on the evaluated delay data; S44. Continue to collect real-time displacement data and re-execute steps S41 to S43; Through the above steps, the influence of the vibration of the device under test on the laser rangefinder is shielded, and the laser rangefinder is kept relatively stable.

[0017] Preferably, in step S41, the training method of the long short-term memory network includes the following steps: S411. Arrange the historical vibration data in chronological order and divide it into a training set and a test set; Preprocess the original data using normalization; S412. Build an LSTM model; S413. Train the model: Use historical data to train the model, and the optimization goal is to reduce the mean square error between the predicted value and the true value; ; Among them, y t is the actual value, is the predicted value, N is the number of samples; S414. Predict future vibration data: Obtain the vibration data at future time points through the above steps.

[0018] Preferably, in step S42, the evaluation of the delay time data between data acquisition and compensation execution adopts the following steps: S421. Decompose the delay time between data acquisition and compensation execution into processing delay, communication delay, and execution delay; The execution delay time is obtained through the system calibration process; S422. Evaluate the sum of the processing delay and the communication delay, that is, the delay time from data acquisition to the output of the control command, using the phase spectrum method model; 1) Perform fast Fourier transform on the input signal x[n] and the output response signal y[n] ; ; 2) Calculate the phase angle at each frequency component; ; 3) For each frequency component k , calculate the phase difference ; ; 4) Estimate the delay time according to the relationship between the phase difference and the frequency: ; where f[k] is the actual frequency value corresponding to the frequency component k ; 5) Select a main frequency component or take the average delay of all valid frequency components as the final estimated value; ; 6) Adjust the obtained actual delay time f s according to the sampling rate (milliseconds); ; Or, in parallel with step S422, S423. Obtain the delay time from data acquisition to control instruction output using an autoregressive model; 1) Let the input signal x[n] and the output signal y[n] be represented by an p -order and an q -order autoregressive model respectively: ; ; where e x [n] and e y [n] are white noise error terms respectively; 2) Use the least squares method or other parameter estimation techniques to determine the autoregressive coefficients a i and b j ; 3) Based on the estimated autoregressive model, solve for the response of the system under a unit impulse input, i.e., the impulse response function h [n] ; 4) Calculate the cross-correlation function x[n] of the input signal h[n] and the impulse response function , and find the peak position ; ; ; 5) According to the signal sampling rate f s Adjust the obtained actual delay time: ; Through the above steps, obtain the sum of the processing delay and the communication delay time, and add the actuator delay time obtained by calibration, that is, obtain the total delay time from data acquisition to compensation output.

[0019] Preferably, step S5 includes the following steps: S51. Filter the vibration data along the x, y, and z axes respectively by using a digital band-pass filter; S52. Perform a fast Fourier transform on the preprocessed data of each axis to obtain a frequency domain model; Calculate the energy of each frequency component as an index for evaluating vibration characteristics; S53. Capture local features in non-stationary signals through wavelet transform, and perform continuous wavelet transform on the data of each axis by using continuous wavelet transform; S54. Combine the data of the three axes into a matrix, where each row represents a three-dimensional vector at a time point, calculate the covariance matrix, and perform eigenvalue decomposition on the covariance matrix; S55. At each time point n , combine the vibration data of the x, y, and z axes into a three-dimensional vector, calculate the modulus length of the combined vector at each time point to evaluate the intensity of the overall vibration; Through the above steps, restore the specific vibration state of the water turbine according to the collected data.

[0020] The present invention provides a water turbine vibration monitoring sensor system and method, which has the following beneficial effects: 1) Set up an actively driven spatial compensation mechanism to overcome the stability problem of the sensor of the conventional stabilizing mechanism in the low-frequency state.

[0021] 2) Use a laser rangefinder to collect the distances to the reflection targets respectively, accurately obtain displacement data, and thus obtain accurate low-frequency vibration data.

[0022] 3) The spatial compensation mechanism adopts a prediction compensation scheme based on artificial intelligence to overcome the vibration caused by delay. Description of the Drawings

[0023] The following further describes the present invention in conjunction with the drawings and embodiments: Figure 1 It is an elevation view of the present invention.

[0024] Figure 2 is Figure 1 the A-A cross-sectional view of

[0025] Figure 3 is Figure 1 the B-B cross-sectional view of

[0026] Figure 4 the control structure block diagram of the present invention

[0027] Figure 5 the overall layout structure schematic diagram of the present invention

[0028] Figure 6 the schematic diagram of the comparison of the monitoring effects between the system of the present invention and the eddy current sensor

[0029] Figure 7 the schematic diagram of the comparison of the monitoring effects between the system of the present invention and other comparison sensors

[0030] Figure 8 the schematic diagram of the amplitude spectrum comparison of the monitoring effects between the system of the present invention and other comparison sensors

[0031] Figure 9 the schematic diagram of the vibration data collected by the system of the present invention under multi-condition conditions

[0032] Figure 10 the partial schematic diagram of the method of the present invention to restore the specific vibration state of the water turbine

[0033] In the figure: vibration sensor 1, inner frame 101, outer frame 102, cantilever opening 103, upper cantilever 104, vertical elastic element 105, base 106, vertical guide post 107, x-direction laser rangefinder 108, sliding sleeve 109, lower cantilever 110, horizontal elastic element 111, sensor top cover 112, x-direction compensation element 113, x-direction hook 114, z-direction laser rangefinder 115, y-direction compensation table 116, barb 117, x-direction compensation table 118, z-direction compensation table 119, z-direction compensation element 120, y-direction hook 121, y-direction laser rangefinder 122, y-direction compensation element 123, water turbine top cover 2, triaxial acceleration sensor 3, reflection target 4, conduction seat 5, main control device 6, host computer 7. Specific embodiments

[0034] Embodiment 1: As Figure 1 , 5 disclosed in, a water turbine vibration monitoring sensor system includes a spatial compensation mechanism, a stabilization mechanism, a ranging sensor, and a reflection target 4; The reflection target 4 is fixedly arranged on the device to be measured. In this example, the device to be measured is the water turbine housing, asFigure 1 As shown in Figure 1 , the reflection target 4 is fixed on the turbine housing with a "cross-shaped" steel structure for transmitting vibration to the reflection target 4; preferably, the reflection target 4 is a regular tetrahedron, and the surface of the reflection target 4 is mirror-treated. In this example, a high-precision mirror reflection target 4 with a side length of 5 mm is used. Figure 1 In Figure 1 , for ease of observation, the volume of the reflection target 4 is locally enlarged.

[0035] The space compensation mechanism is connected to the device under test, and the stabilization mechanism is arranged on the space compensation mechanism. The space compensation mechanism is used to compensate for the influence of vibration on the stabilization mechanism; the space compensation mechanism refers to a mechanism for actively compensating for displacements in the x-axis, y-axis, and z-axis caused by vibration. The stabilization mechanism refers to a mechanism for passively compensating for displacements in the x-axis, y-axis, and z-axis caused by vibration.

[0036] The stabilization mechanism is used to maintain the relative stability of the spatial position. The distance measurement sensor is arranged inside the stabilization mechanism. The distance measurement sensor detects the distance data of the spatial component between it and the reflection target 4 to obtain the low-frequency vibration data of the device under test. In this example, the distance measurement sensor is preferably a laser distance meter, such as Keyence CL-L(P)007G, which has a maximum resolution of up to 0.015 μm under high-precision working conditions, meeting the accuracy requirements.

[0037] The preferred solution is as Figure 1 shown in Figure 1 , the space compensation mechanism includes an x-direction compensation table 118, a y-direction compensation table 116, and a z-direction compensation table 119 that are stacked on top of each other; Compensation elements are arranged between the x-direction compensation table 118, the y-direction compensation table 116, and the z-direction compensation table 119, and the relative movement between the compensation tables is driven by the compensation elements. That is, the relative movement between the compensation tables along the x-axis, y-axis, and z-axis.

[0038] In the preferred solution, in the space compensation mechanism, the base 106 is fixedly connected to the device under test. The x-direction compensation table 118 is slidably arranged horizontally on the base 106. Preferably, a slide rail in the x-direction is arranged between the x-direction compensation table 118 and the base 106. An x-direction hook 114 extending towards the base 106 is arranged on the x-direction compensation table 118 along the x-direction. An x-direction compensation element 113 is arranged between the x-direction hook 114 and the base 106; to drive the x-direction compensation table 118 to reciprocate along the x-axis, as Figure 3 shown in Figure 3 . Figure 3 Figure 3 shows a solution driven unidirectionally. This structure is more simplified, and the accuracy interference generated by the system due to the structure is extremely low.

[0039] The y-direction compensation stage 116 is horizontally slidably arranged on the x-direction compensation stage 118. Preferably, a slide rail in the y-direction is provided between the y-direction compensation stage 116 and the x-direction compensation stage 118. A y-direction hook 121 extending towards the x-direction compensation stage 118 is provided along the y-direction on the y-direction compensation stage 116. A y-direction compensation element 123 is provided between the y-direction hook 121 and the x-direction compensation stage 118; to drive the y-direction compensation stage 116 to reciprocate along the y-axis.

[0040] The z-direction compensation stage 119 is arranged on the y-direction compensation stage 116. A z-direction compensation element 120 is provided between the z-direction compensation stage 119 and the y-direction compensation stage 116; to drive the z-direction compensation stage 119 to reciprocate along the Z-axis.

[0041] The compensation element includes a piezoelectric element or a linear motor. In this example, a piezoelectric element is preferably used, that is, the inverse piezoelectric effect of the piezoelectric element is utilized in this example. When an electric field is applied to the piezoelectric element, the piezoelectric element will deform, thereby driving the movement of the compensation stage.

[0042] The preferred solution is as Figure 3 In, the compensation element is a multi-stack piezoelectric element, and each piezoelectric element is independently driven by a driving device. That is, each piezoelectric element has a separate electrode to independently supply power to the piezoelectric element to drive the movement of the piezoelectric element. Each piezoelectric element is polarized in the same direction. By controlling the power supply to the independent piezoelectric element, it is also used to achieve precise control of the displacement amount.

[0043] The preferred solution is as Figure 1 、 2 In, the stabilizing mechanism includes an inner frame 101 and an outer frame 102; The bottom of the outer frame 102 is horizontally slidably arranged on the top of the z-direction compensation stage 119. No guide rail is provided between the outer frame 102 and the z-direction compensation stage 119, so that the outer frame 102 can have an unrestricted relative displacement along the horizontal direction. When the water turbine housing vibrates, affected by inertia, the outer frame 102 can filter the horizontal movement of the water turbine housing. However, under low-frequency vibration conditions, due to the existence of friction, that is, the existence of static friction, the low-frequency horizontal displacement may not be filtered by the outer frame 102. Therefore, the provided spatial compensation mechanism is mainly used to assist the outer frame 102 to better filter the horizontal vibration through displacement compensation. A plurality of horizontally arranged horizontal elastic elements 111 are provided along the circumference between the z-direction compensation stage 119 and the outer frame 102. The plurality of horizontal elastic elements 111 are balanced and align the outer frame 102 and the z-direction compensation stage 119 in the center; the horizontal elastic elements 111 are used to make the outer frame 102 and the z-direction compensation stage 119 tend to be centered and aligned.

[0044] The inner frame 101 is slidably connected to the outer frame 102 in the vertical direction. A plurality of vertical elastic elements 105 are arranged along the circumference between the inner frame 101 and the outer frame 102, and the self-weight of the vertical elastic elements 105 and the inner frame 101 constitutes a balance. With this structure, when vibration occurs in the vertical direction, the vertical displacement is filtered by the relative sliding between the inner frame 101 and the outer frame 102. Similarly, under the condition of low-frequency vibration, due to the existence of friction, that is, static friction, the low-frequency horizontal displacement may be difficult to be filtered by the outer frame 102. The active displacement of the z-direction compensation platform 119 of the space compensation mechanism assists the inner frame 101 to better filter the vertical vibration.

[0045] As Figure 1 , 2 shown in, at least three laser rangefinders respectively point to the reflection target 4 in the x-direction, y-direction and z-direction. The reflection target 4 is provided with corresponding vertical reflection surfaces. The laser beam emitted by the laser rangefinder is reflected by the reflection target 4 and then received by the laser rangefinder. According to the time difference between emission and reception, the distance between the laser rangefinder and the reflection target 4 is obtained. Due to the setting of the space compensation mechanism and the stabilization mechanism, the inner frame 101 is kept relatively stationary in a passive and active manner, and is hardly affected by the vibration of the water turbine housing. The vibration of the water turbine housing is transmitted to the reflection target 4. Therefore, the vibration data of the water turbine housing can be obtained from the distance between the laser rangefinder and the reflection target 4. Combining with the time parameter, the three-dimensional vibration situation in the time domain of the water turbine housing can be restored.

[0046] As Figure 1 shown in, a triaxial acceleration sensor 3 is provided on the z-direction compensation platform 119. The set triaxial acceleration sensor 3 adopts, for example, a capacitive acceleration sensor such as Murata MA402, NXP FXLS8971CF or Vite Intelligent Xsens accelerometer gyroscope. It is used to detect the vibration situation after compensation, and control the action of the compensation element according to the collected vibration data to actively compensate for the displacement generated by the vibration. In this example, a threshold is set in the collected vibration data for compensation. For example, the threshold is 10~100Hz. Vibration higher than the threshold, such as 100Hz, is not actively compensated because it can be absorbed by the stabilization mechanism, while vibration lower than 100Hz is actively compensated. So that the overall amplitude of the vibration collected by the triaxial acceleration sensor 3 shows a decreasing trend.

[0047] The preferred solution is as Figures 1 - 3Among them, the outer frame 102 is a cavity structure. There is an opening at the center of the outer frame 102. A plurality of barbs 117 are arranged along the circumference on the z-direction compensation table 119. The barbs 117 pass through the opening. A horizontal elastic element 111 is arranged between the barbs 117 and the outer frame 102; preferably, both the horizontal elastic element 111 and the vertical elastic element 105 are high-precision springs. What is not shown in the figure is that in order to ensure the balance of the tension, a tension adjustment mechanism is arranged between the horizontal elastic element 111 and the compensation table. For example, the horizontal elastic element 111 is fixedly connected to a bolt, and the bolt passes through the barb 117 and is connected to a nut. Adjusting the nut can precisely adjust the tension between the horizontal elastic elements 111, thereby ensuring that a plurality of horizontal elastic elements 111 form a balance.

[0048] Vertical guide posts 107 are arranged along the circumference on the inner wall of the outer frame 102. Corresponding sliding sleeves 109 are arranged on the inner frame 101. The sliding sleeves 109 are slidably connected to the vertical guide posts 107; A plurality of cantilever openings 103 are arranged along the circumference on the side wall of the inner frame 101. Corresponding upper cantilevers 104 are arranged along the circumference on the outer frame 102. The upper cantilevers 104 pass through the cantilever openings 103 and extend into the inner frame 101. Lower cantilevers 110 are arranged along the circumference on the inner frame 101. The vertical elastic element 105 is arranged between the upper cantilevers 104 and the lower cantilevers 110. In this example, a hanging structure is adopted, and the force balance can be conveniently achieved by adjusting the tension of the vertical elastic element 105. What is not shown in the figure is that the tension adjustment is achieved by adjusting the screw connected to the vertical elastic element 105. The vertical elastic element 105 preferably adopts a tension spring.

[0049] The preferred solution is as Figure 1 、 2 Among them, the reflection target 4 is fixedly connected to the device under test through the conduction seat 5. The x-direction laser rangefinder 108 and the y-direction laser rangefinder 122 are horizontally arranged. The x-direction laser rangefinder 108 and the y-direction laser rangefinder 122 are 90° apart on the circumference. The x-direction laser rangefinder 108 and the y-direction laser rangefinder 122 respectively vertically point to two vertical planes of the reflection target 4 that are perpendicular to each other; The z-direction laser rangefinder 115 is arranged at the bottom of the sensor top cover 112 of the inner frame 101. The z-direction laser rangefinder 115 vertically points to the horizontal top surface of the reflection target 4.

[0050] In the preferred solution, a magnetic levitation mechanism is arranged between the bottom of the outer frame 102 and the top of the z-direction compensation table 119. The magnetic levitation mechanism is an array of magnets with opposite polarities; with this structure, the outer frame 102 can better filter the vibration displacement, especially the displacement generated by vibrations above 100 Hz.

[0051] In the preferred solution, a counterweight is arranged in the inner frame 101 to make the center of mass of the inner frame 101 coincide with the central axis of the inner frame 101. The counterweight mainly compensates for the weight of the laser rangefinder.

[0052] In a preferred embodiment, a vibration sensor is provided inside the reflection target 4 to collect the vibration data of the device under test. The vibration sensor uses a triaxial vibration sensor, which is mainly used to collect the vibration data of the reflection target 4 above 50 Hz to compensate for the problem of low high-frequency vibration accuracy of the sensor in this example.

[0053] A preferred embodiment is as Figure 4 described, a main control device 6 and a host computer 7 are also provided; the main control device 6 in this example uses a PLC, and the host computer is an industrial computer or a server.

[0054] The triaxial acceleration sensor 3 is electrically connected to the main control device 6 to collect the spatial displacement of the z-direction compensation table 119. The main control device 6 is electrically connected to the compensation element to compensate for the influence of vibration on the spatial position of the z-direction compensation table 119; although the triaxial acceleration sensor 3 has the problem of cumulative error, the compensation in the present invention occurs within a time range of milliseconds, so the disadvantage of high cumulative error of the triaxial acceleration sensor is avoided.

[0055] The laser rangefinder sends the collected data to the host computer 7 through the main control device 6 for analyzing the vibration data. The ranging data is converted into three-dimensional space data and combined with the time axis to obtain the water turbine housing as Figure 6 described, in a certain water turbine housing, eddy current sensors, namely ECT sensors, are respectively arranged, Figure 6 the data in the first row at the bottom in Figure 6 and the system of the present invention,

[0056] as Figure 7 described, the system of the present invention is also compared with more sensors, and it is shown that at low frequencies, larger amplitude changes can be collected, and better monitoring accuracy is achieved at low frequencies. Figure 7 In Figure 7 described, HKTA is a Hong Kong Fujitsu horizontal optoelectronic sensor, GZJX is a Shenzhen Anrui horizontal ultrasonic sensor, QHKJ is a Shenzhen Shenpu horizontal mechanical sensor, and AJY is a Shanghai Keyence horizontal laser sensor without active compensation for detection. As Figure 9 described, in the amplitude spectrum diagram, the system of the present invention also has a high amplitude progress in the low-frequency band. Moreover, the waveform accuracy is high, and the eddy current sensor and the AJY horizontal sensor without active compensation have incorrect peaks in the frequency band of about 0.25 Hz.

[0057] Embodiment 2: A monitoring method using the above water turbine vibration monitoring sensor system includes the following steps: S1. When the device under test is in a shutdown state, set the ranging data between the laser rangefinder and the reflection target 4 to zero; S2. Start the device under test, and the laser rangefinder collects the distance data between it and the reflection target 4; S3. A triaxial acceleration sensor 3 is provided on the z-direction compensation table 119, and the acceleration data of the inner frame 101 is obtained and converted into real-time displacement data; S4. There is also a main control device 6. The main control device 6 decomposes the real-time displacement data into x-direction, y-direction, and z-direction displacement data, and respectively controls the x-direction compensation element 113, y-direction compensation element 123, and z-direction compensation element 120 to move with opposite displacement amounts in real time to compensate for the influence of vibration on the spatial position of the inner frame 101; The vibration compensation is based on the vibration data detected by the triaxial vibration sensor. The problem of the triaxial vibration sensor is that the cumulative error is relatively high, making the final monitoring data not referenceable. However, during the vibration process at the millisecond level, due to the low cumulative error, it does not affect the acquisition accuracy of the vibration sensor. The triaxial vibration sensor can feedback relatively accurate vibration data to guide high-precision compensation operations.

[0058] S5. The main control device 6 stores the ranging data and time data of the laser rangefinder; The monitoring of the low-frequency vibration of the water turbine is realized through the above steps.

[0059] Preferably, in step S4, it includes the following steps: S41. Predict future vibration data using a long short-term memory network; S42. Evaluate the delay time data between data acquisition and compensation execution of the system using the phase spectrum method or the autoregressive model method; S43. Perform compensation according to the predicted future vibration data based on the evaluated delay data; S44. Continue to collect real-time displacement data and re-execute steps S41 to S43; Through the above steps, the influence of the vibration of the device under test on the laser rangefinder is shielded, and the laser rangefinder is kept relatively stable.

[0060] Preferably, in step S41, the training method of the long short-term memory network includes the following steps: S411. Arrange the historical vibration data in chronological order and divide it into a training set and a test set; Preprocess the original data using normalization; S412. Construct an LSTM model; Due to the non - linear characteristics of vibration, a long short - term memory network (LSTM) is used for prediction; the method includes the following steps: 1) The LSTM cell contains an input gate, a forget gate, and an output gate. These gates control the flow of information and solve the problem of gradient disappearance in RNNs; Let there be an LSTM cell, and the state update formula at time step t is as follows: 2) Forget gate: Decides which information to discard from the cell state; (1); 3) Input gate: Decides which new information will be added to the cell state; (2); 4) Candidate cell state: (3); 5) Cell state update: (4); 6) Output gate: Decides the next hidden state; (5); 7) Hidden state update: (6); Among them, W f 、W i 、W C 、W o are weight matrices, b f 、b i 、b C 、b o are bias terms, is an activation function, indicating that the input is mapped to the interval (0, 1) by the sigmoid function. The role of the sigmoid function is to convert the result of the linear combination into a probability value, which is used to decide which information should be retained or forgotten; tanh represents the hyperbolic tangent activation function, is the output of the forget gate, which is a vector with values between 0 and 1. It represents which information to discard from the cell state at time step t and is generated by the sigmoid function. A value close to 0 means "discard", and a value close to 1 means "retain"; h t−1 is the hidden state vector of the previous time step. The hidden state contains information from previous time steps and is passed to the current time step so that historical information can be considered when processing the current input;x t is the input vector at the current time step. For the vibration prediction of the water turbine unit, it is the vibration data at a certain moment; i t is the output of the input gate, indicating that at the time step t decides which new information will be added to the cell state; represents the candidate cell state at the current time step t; represents the cell state at the current time step t; is the output of the output gate; is the hidden state at the current time step t.

[0061] S413. Training the model: Use historical data to train the model, and the optimization goal is to minimize the mean square error between the predicted value and the true value; 7; wherein, y t is the actual value, is the predicted value, N is the number of samples; S414. Predicting future vibration data: Obtain the vibration data at future time points through the above steps.

[0062] Preferably, in step S42, the following steps are adopted for the data evaluation of the delay time between data acquisition and execution of compensation: S421. In the control system, the total delay time from data acquisition to compensation output can be decomposed into the sum of multiple components, including processing delay, communication delay, and execution delay; (8); : The total delay time from data acquisition to compensation output; : Is the data processing delay time of the controller; respectively represent the processing time at the sending end, the network transmission time, and the processing time at the receiving end, and their sum is the communication delay; : The execution mechanism delay time. The execution delay time is obtained through the system calibration process; The method of system calibration is to send instructions to the execution mechanism, in this example, the compensation element, such as the x-direction compensation element 113, the y-direction compensation element 123, and the z-direction compensation element 120, and use a high-precision displacement sensor, such as a grating displacement sensor, to detect the final displacement distance. The time from the instruction sending time point to the end of the displacement detection is the execution delay time.

[0063] S422. Evaluate the sum of the processing delay and the communication delay using the phase spectrum method model, i.e., the delay time from data acquisition to the output of the control command; 1) For the input signal x[n] and the output response signal y[n] perform a fast Fourier transform; (9); 2) Calculate the phase angle at each frequency component; (10); 3) For each frequency component k , calculate the phase difference ; (11); 4) Estimate the delay time based on the relationship between the phase difference and the frequency: (12); where f[k] is the actual frequency value corresponding to the frequency component k ; 5) Select a main frequency component or take the average delay of all valid frequency components as the final estimated value; (13); 6) Adjust the obtained actual delay time f s according to the sampling rate (milliseconds); (14); Alternatively, parallel to step S422, S423. Obtain the delay time (milliseconds) from data acquisition to the output of the control command using an autoregressive model; 1) Assume that the input signal x[n] and the output signal y[n] are represented by p -order and q -order autoregressive models respectively: ; (15); where e x [n] and e y [n] are white noise error terms respectively; 2) Use the least squares method or other parameter estimation techniques to determine the autoregressive coefficientsa i and b j ; 3) Based on the estimated autoregressive model, solve for the response of the system under a unit impulse input, i.e., the impulse response function h [n] ; 4) Calculate the input signal x[n] and the impulse response function h[n] to obtain the cross-correlation function , and find the peak position ; (16); (17); 5) Adjust the obtained actual delay time according to the signal sampling rate f s : (18); By the above steps, obtain the sum of the processing delay and the communication delay time, and add the actuator delay time obtained by calibration, i.e., obtain the total delay time from data acquisition to compensation output.

[0064] Preferably, step S5 includes the following steps: S51. Filter the vibration data along the x, y, and z axes respectively using a digital band-pass filter; The specific steps are as follows: Set a band-pass filter with a low cut-off frequency of f low , a high cut-off frequency of f high , and a sampling frequency of fs ; Calculate the normalized frequency: (19); Normalize to the range [0, 1]: (20); The transfer function of the digital band-pass filter is: (21); where the coefficients b i and a j are calculated using the `cheby1` function in the SciPy library; Apply the filtering function to the vibration data along the x, y, and z axes respectively: (22); Through the above steps, the vibration data in the x, y, and z directions after band-pass filtering are obtained.

[0065] S52. Perform a fast Fourier transform on the preprocessed data for each axis to obtain a frequency-domain model; The specific steps are as follows: Fast Fourier transform; Perform a fast Fourier transform on the preprocessed data for each axis to obtain a frequency-domain representation: (23); Calculate the power spectral density; Calculate the energy of each frequency component as an index for evaluating vibration characteristics: (24); Calculate the energy of each frequency component as an index for evaluating vibration characteristics.

[0066] S53. Capture local features in non-stationary signals through wavelet transform. Use continuous wavelet transform to perform continuous wavelet transform on the data for each axis; (25); ; ; (26); Among them, is the mother wavelet function, a is the scale parameter that controls the width of the wavelet, b is the translation parameter that controls the position of the wavelet.

[0067] S54. Combine the data of the three axes into a matrix. Each row represents a three-dimensional vector at a time point. Calculate the covariance matrix and perform eigenvalue decomposition on the covariance matrix; Calculate the covariance matrix Combine the data of the three axes into a matrix , and each row represents a three-dimensional vector at a time point: (27); Calculate the covariance matrix : (28); Eigenvalue decomposition. Perform eigenvalue decomposition on the covariance matrix: (29); Among them, is the eigenvector matrix, is the eigenvalue matrix, and the elements on the diagonal are eigenvalues; Select the main components, and select the first k main components according to the eigenvalue size: (30); S55. At each time point n , synthesize the vibration data of the x, y, and z axes into a three-dimensional vector , and calculate the modulus of the synthesized vector at each time point to evaluate the intensity of the overall vibration; At each time point n , synthesize the vibration data of the three axes into a three-dimensional vector: (31); Calculate the modulus of the synthesized vector at each time point to evaluate the intensity of the overall vibration: (32); Through the above steps, restore the specific vibration state of the water turbine according to the collected data. As Figure 10 shown in Figure 10 , the schematic diagram of the elliptical trajectory formed by the vibration is shown in the upper right corner.

[0068] The above embodiments are only the preferred technical solutions of the present invention and should not be regarded as limitations on the present invention. The embodiments and the features in the embodiments in this application can be arbitrarily combined with each other without conflict. The protection scope of the present invention should be the technical solutions recorded in the claims, including the equivalent replacement solutions of the technical features in the technical solutions recorded in the claims. That is, the equivalent replacement improvements within this scope are also within the protection scope of the present invention.

Claims

1. A turbine vibration monitoring sensor system, characterized in that: it includes a space Compensation mechanism, stabilization mechanism, distance sensor and reflective target (4); The reflective target (4) is fixed on the device under test and is used to transmit vibration; The space compensation mechanism is connected to the device under test, and the stabilizing mechanism is arranged on the space compensation mechanism, and the space compensation mechanism is used to compensate for the influence of vibration on the stabilizing mechanism; The stabilizing mechanism is used to maintain the relative stability of the spatial position. The distance measuring sensor is arranged in the stabilizing mechanism. The distance measuring sensor detects the distance data of the spatial component between the distance measuring sensor and the reflective target (4) to obtain the low-frequency vibration data of the device under test.

2. A turbine vibration monitoring sensor system according to claim 1, characterized in that: The compensation mechanism comprises an x-direction compensation platform (118), a y-direction compensation platform (116), and a z-direction compensation platform (119) stacked on each other; Compensation elements are provided between the x-direction compensation platform (118), the y-direction compensation platform (116) and the z-direction compensation platform (119), and the compensation elements drive the compensation platforms to move relative to each other.

3. A turbine vibration monitoring sensor system according to claim 1, characterized in that: In the compensation mechanism, a base (106) is fixedly connected to a device under test, an x-direction compensation platform (118) is horizontally slidably arranged on the base (106), an x-direction hook (114) extending toward the base (106) is arranged on the x-direction compensation platform (118) along the x-direction, and an x-direction compensation element (113) is arranged between the x-direction hook (114) and the base (106); The y-direction compensation platform (116) is horizontally slidably arranged on the x-direction compensation platform (118); a y-direction hook (121) extending toward the x-direction compensation platform (118) is arranged on the y-direction compensation platform (116) along the y-direction; and a y-direction compensation element (123) is arranged between the y-direction hook (121) and the x-direction compensation platform (118); The z-direction compensation platform (119) is arranged on the y-direction compensation platform (116), and a z-direction compensation element (120) is arranged between the z-direction compensation platform (119) and the y-direction compensation platform (116); Compensation elements include piezoelectric elements or linear motors.

4. A turbine vibration monitoring sensor system according to claim 3, characterized in that: The compensation element is a plurality of stacked piezoelectric elements, and each piezoelectric element is independently driven by a driving device.

5. A turbine vibration monitoring sensor system according to claim 3 or 4, characterized in that: The mechanism comprises an inner frame (101) and an outer frame (102); The outer frame (102) is horizontally slidably arranged on the top of the z-direction compensation platform (119), and a plurality of horizontally arranged horizontal elastic elements (111) are provided along the circumference between the z-direction compensation platform (119) and the outer frame (102), wherein the plurality of horizontal elastic elements (111) form a balance and centrally align the outer frame (102) and the z-direction compensation platform (119); The inner frame (101) and the outer frame (102) are slidably connected in the vertical direction, and a plurality of vertical elastic elements (105) are provided along the circumference between the inner frame (101) and the outer frame (102), and the vertical elastic elements (105) and the self-weight of the inner frame (101) form a balance; At least three laser rangefinders are respectively pointed toward the reflective target (4) along the x-direction, the y-direction and the z-direction, and the reflective target (4) is provided with a corresponding vertical reflective surface; A three-axis acceleration sensor (3) is provided on the z-direction compensation platform (119).

6. A turbine vibration monitoring sensor system according to claim 5, characterized in that: The outer frame (102) is a hollow structure, an opening is provided at the center of the outer frame (102), a plurality of barbs (117) are provided along the circumference of the z-direction compensation platform (119), the barbs (117) pass through the opening, and a horizontal elastic element (111) is provided between the barbs (117) and the outer frame (102); A vertical guide column (107) is provided along the circumference of the inner wall of the outer frame (102), and a corresponding sliding sleeve (109) is provided on the inner frame (101), wherein the sliding sleeve (109) is slidably connected to the vertical guide column (107); A plurality of cantilever openings (103) are provided along the circumference of the side wall of the inner frame (101); a corresponding upper cantilever (104) is provided along the circumference of the outer frame (102); the upper cantilever (104) passes through the cantilever opening (103) and extends into the inner frame (101); a lower cantilever (110) is provided along the circumference of the inner frame (101); and a vertical elastic element (105) is arranged between the upper cantilever (104) and the lower cantilever (110).

7. A turbine vibration monitoring sensor system according to claim 6, characterized in that: The reflective target (4) is fixedly connected to the device to be measured via a conductive seat (5); an x-direction laser rangefinder (108) and a y-direction laser rangefinder (122) are arranged horizontally; the x-direction laser rangefinder (108) and the y-direction laser rangefinder (122) are 90° apart on a circle; the x-direction laser rangefinder (108) and the y-direction laser rangefinder (122) are respectively pointed vertically to two mutually perpendicular vertical surfaces of the reflective target (4); A z-direction laser rangefinder (115) is arranged at the bottom of a sensor top cover (112) of the inner frame (101), and the z-direction laser rangefinder (115) is vertically pointed toward the horizontal top surface of the reflective target (4).

8. A turbine vibration monitoring sensor system according to claim 5, characterized in that: A magnetic suspension mechanism is provided between the bottom of the outer frame (102) and the top of the z-direction compensation platform (119), and the magnetic suspension mechanism is a magnet array with opposite polarities.

9. A turbine vibration monitoring sensor system according to claim 8, characterized in that: A counterweight is provided on the inner frame (101) so that the center of mass of the inner frame (101) coincides with the central axis of the inner frame (101).

10. A turbine vibration monitoring sensor system according to claim 9, characterized in that: A vibration sensor is arranged inside the reflective target (4) to collect vibration data of the device under test.

11. A turbine vibration monitoring sensor system according to any one of claims 1 to 10, characterized in that: A main control device (6) and a host computer (7) are also provided; The triaxial acceleration sensor (3) is electrically connected to the main control device (6) to collect the spatial displacement of the z-direction compensation platform (119), and the main control device (6) is electrically connected to the compensation element to compensate for the influence of vibration on the spatial position of the z-direction compensation platform (119); The laser rangefinder sends the collected data to the host computer (7) via the main control device (6) for analyzing the vibration data.

12. A monitoring method using a turbine vibration monitoring sensor system according to any one of claims 1 to 11, characterized in that: The following steps are involved: S1, when the device under test is in a stopped state, setting the distance measurement data between the laser rangefinder and the reflective target (4) to zero; S2, starting the device under test, and the laser rangefinder collects distance data between the device and the reflective target (4); S3, a three-axis acceleration sensor (3) is provided on the z-direction compensation platform (119) to obtain acceleration data of the inner frame (101) and convert it into real-time displacement data; S4, a main control device (6) is also provided, the main control device (6) decomposes the real-time displacement data into x-direction, y-direction and z-direction displacement data, and controls the x-direction compensation element (113), the y-direction compensation element (123) and the z-direction compensation element (120) to move in opposite displacement amounts in real time, so as to compensate for the influence of the vibration on the spatial position of the inner frame (101); S5, the main control device (6) stores the distance data and time data of the laser rangefinder; The above steps can be used to monitor the low-frequency vibration of the turbine.

13. A monitoring method for a hydraulic turbine vibration monitoring sensor system according to claim 12, The method is characterized in that step S4 comprises the following steps: S41. Predicting future vibration data using long short-term memory networks; S42, evaluating the delay time data of the system from data collection to compensation execution by using a phase spectrum method or an autoregressive model method; S43, performing compensation according to the estimated delay data and the predicted future vibration data; S44, continue to collect real-time displacement data, and re-execute steps S41 to S43; Through the above steps, the influence of the vibration of the device under test on the laser rangefinder is shielded, so that the laser rangefinder remains relatively stable.

14. A monitoring method for a water turbine vibration monitoring sensor system according to claim 13, characterized in that In step S41, the training method of the long short-term memory network includes the following steps: S411, arranging the historical vibration data in chronological order, and dividing the historical vibration data into a training set and a test set; Preprocess the raw data using normalization; S412, build LSTM model; S413, training model: using historical data to train the model, the optimization goal is to reduce the mean square error between the predicted value and the true value; ； in, y t is the actual value, is the predicted value, N is the sample size; S414, predict future vibration data: Through the above steps, the vibration data at the future time point is obtained.

15. The monitoring method of a hydraulic turbine vibration monitoring sensor system according to claim 13, characterized in that In step S42, the delay time data between data collection and compensation execution is evaluated by the following steps: S421, decomposing the delay time between data collection and compensation execution into processing delay, communication delay and execution delay; The execution delay time is obtained through the system calibration process; S422, using a phase spectrum method model to evaluate the sum of processing delay and communication delay, that is, the delay time from data acquisition to control command output; 1) Input signal x[n] and output response signal y[n] Perform a fast Fourier transform; ； 2) Calculate the phase angle of each frequency component; ； 3) For each frequency component k , calculate the phase difference ; ； 4) Estimate the delay time based on the relationship between phase difference and frequency: ； in, f[k] is the frequency component corresponding to k The actual frequency value of 5) Select a major frequency component Or take all effective frequency components Average delay As the final estimate; ； 6) According to the sampling rate f s Adjust the actual delay time (millisecond); ； Alternatively, in parallel with step S422, S423, using an autoregressive model to obtain a delay time from data collection to control command output; 1) Set the input signal x[n] and output signal y[n] Use p Step and q Order autoregressive model: ； ； in, e x [n] and e y [n] are the white noise error terms; 2) Determine the autoregressive coefficients using the least squares method or other parameter estimation techniques a i and b j ; 3) Based on the estimated autoregressive model, solve the response of the system under unit pulse input, that is, the impulse response function h[n] ; 4) Calculate the input signal x[n] With impulse response function h[n] The cross-correlation function , find the peak position ; ； ； 5) According to the signal sampling rate f s The actual delay time obtained by adjustment: ； Through the above steps, the sum of the processing delay and the communication delay time is obtained, and the total delay time from data acquisition to compensation output is obtained by adding the calibrated actuator delay time.

16. The monitoring method of a hydraulic turbine vibration monitoring sensor system according to claim 12, wherein step S5 comprises the following steps: S51, respectively filtering the vibration data along the x, y, and z axes using a digital bandpass filter; S52, performing fast Fourier transform on the preprocessed data of each axis to obtain a frequency domain model; Calculate the energy of each frequency component as an indicator for evaluating vibration characteristics; S53, capturing local features in the non-stationary signal by wavelet transform, and performing continuous wavelet transform on the data of each axis by using continuous wavelet transform; S54, merging the data of the three axes into a matrix, where each row represents a three-dimensional vector at a time point, calculating the covariance matrix, and performing eigenvalue decomposition on the covariance matrix; S55. At each time point n , synthesize the vibration data of the x, y, and z axes into a three-dimensional vector, and calculate the modulus of the synthetic vector at each time point to evaluate the intensity of the overall vibration; Through the above steps, the specific vibration state of the turbine is restored based on the collected data.

Images