A method and system for eliminating crosstalk and dynamic compensation of laser vibration measurement of a patrol robot

Abstract

The application discloses a kind of laser vibration measurement technology field, and a kind of laser vibration measurement technology field is eliminated and dynamic compensation method and system of patrol robot laser vibration measurement crosstalk, belongs to.It is with variable stiffness active vibration isolation module, according to motion state signal switches between rigid lock and flexible vibration isolation state;When measuring, the vibration crosstalk component is deducted from original measurement signal using three-axis acceleration sensor to collect laser Doppler vibration measuring instrument LDV host vibration, with adaptive filtering;At the same time, by polarization beam splitter and array detector constitute polarization-space double diversity receiving module, calculate each channel interference signal carrier-to-noise ratio and weighted fusion, the high signal-to-noise ratio signal after fusion is used as the input original measurement signal of adaptive filtering.Physical vibration isolation and algorithm compensation eliminate full-band vibration crosstalk together, double diversity and dynamic fusion realize efficient speckle suppression under compact structure, for high-precision vibration measurement of target object in the progress of patrol robot.

Description

Technical Field

[0001] This invention belongs to the field of laser vibration measurement technology, and in particular to a method and system for vibration interference resistance and noise suppression in laser Doppler vibration measurement for mobile robot platforms. Background Technology

[0002] Inspection robots equipped with laser Doppler vibration meters for in-motion vibration measurement represent an important technological direction for the automated inspection of target structures such as power equipment, bridges, and pipelines. Airborne laser Doppler vibration meters face two core technical challenges in mobile measurements: platform vibration crosstalk and speckle noise, both of which jointly restrict the improvement of measurement accuracy.

[0003] Regarding platform vibration crosstalk, laser Doppler vibration meters measure the relative vibration between the target object and the laser head. When a laser Doppler vibration meter is installed on an inspection robot, the robot's own vibration is transmitted to the laser head through the mounting base. This vibration is superimposed on the measurement signal, forming vibration crosstalk. To suppress this crosstalk, existing solutions typically use passive vibration isolation elements such as rubber vibration isolation pads or spring vibration isolators between the laser Doppler vibration meter and the mounting base. The vibration isolation characteristics are determined by the fixed stiffness and damping of these elements and cannot be adjusted according to the robot's movement or measurement status. Another approach uses piezoelectric actuators or voice coil motors combined with force sensors to form a closed-loop active vibration isolation system. The controller controls the actuator to apply a reverse compensation force based on force sensor feedback. This approach can achieve high vibration isolation accuracy under fixed operating conditions, but its vibration isolation stiffness is fixed and cannot switch between rigid connection and flexible vibration isolation states. When the robot is moving, insufficient rigidity causes the laser Doppler vibration meter to shake, affecting motion stability; when the robot is stationary and measuring, insufficient flexibility causes base vibration to be transmitted into the measurement optical path, limiting vibration isolation efficiency. This contradiction stems from the diametrically opposed requirements of the vibration isolation system's stiffness for the two working conditions of walking and measuring. The mechanical boundaries of existing active vibration isolation systems are fixed at the initial design stage, and their equivalent stiffness characteristics cannot be fundamentally changed by adjusting control parameters.

[0004] Another approach involves adding a reflector and beam splitter to the laser Doppler vibration meter for post-vibration measurement and algorithm correction of the instrument itself. However, this approach is offline, limiting its real-time performance, and it doesn't coordinate with the front-end vibration isolation system, failing to provide dynamic compensation simultaneously with vibration occurrence. Furthermore, even existing vibration isolation schemes with switchable states don't consider the impact of isolation state switching on subsequent vibration compensation algorithms. When the system is in a rigidly locked, non-measurement state, the vibration sensor's reading characteristics differ significantly from those in a flexible isolation state. If the compensation algorithm runs continuously without differentiation, it can easily lead to erroneous convergence of the algorithm model and parameter misalignment, thus affecting the compensation accuracy and response speed when switching back to the measurement state.

[0005] Regarding speckle noise, when a laser irradiates a rough surface, the scattered light at each point on the surface is randomly phase-distributed, forming a speckle pattern in space. Minor vibrations of the robot platform cause the laser spot to move in-plane on the target surface, generating random phase modulation and reducing the signal-to-noise ratio of the interference signal. Existing suppression schemes mainly employ diversity techniques. Spatial diversity schemes typically use multiple independent probes to receive scattered light from different spatial locations, resulting in a large system structure that is not conducive to airborne integration. Polarization diversity schemes achieve signal diversity by simultaneously receiving two orthogonal polarization components in the return light, but they only have two channels and typically require complex optical components such as Faraday rotators in the optical path, leading to a complex structure and high cost. Due to the randomness and variability of speckle noise, a single diversity method presents a difficult trade-off between the number of channels, structural compactness, and noise suppression robustness. Under conditions of local depolarization or deep fading, the signal quality of the remaining channels cannot be guaranteed. Summary of the Invention

[0006] To address the aforementioned issues, this invention provides a method and system for eliminating and dynamically compensating for crosstalk in laser vibration measurement of inspection robots. This invention employs a variable stiffness active vibration isolation module, dynamically switching between rigid locking and flexible vibration isolation states based on motion state signals. Simultaneously, a triaxial accelerometer collects the vibration of the LDV host itself. An adaptive filtering algorithm subtracts vibration crosstalk components from the original measurement signal, constructing a full-band vibration suppression chain that combines physical isolation and algorithmic compensation. Furthermore, a polarization-space dual diversity receiving module is constructed using a polarization beamsplitter and an array detector to calculate the carrier-to-noise ratio of interference signals from each channel and perform weighted fusion. The fused measurement signal is iteratively used as the original measurement signal, forming a full-link signal processing architecture that integrates optical noise suppression and electrical crosstalk compensation. This overcomes the inherent contradiction in airborne laser vibration measurement—where walking stability and measurement vibration isolation accuracy cannot be simultaneously achieved, and the speckle suppression capability of a single diversity method is limited—and enables high-precision vibration measurement of targets by inspection robots during movement.

[0007] To achieve the above objectives, the technical solution adopted by the present invention is: a method for eliminating vibration crosstalk and dynamically compensating for laser vibration measurement during the movement of an inspection robot, comprising: acquiring the motion state signal of the inspection robot; A variable stiffness active vibration isolation module is adopted. According to the motion state signal, it switches to a rigid locking state in the moving state, so that the LDV host and the robot base form a rigid integrated structure. In the measurement state, it switches to a flexible vibration isolation state. In the measurement state, the initial raw measurement signal of the LDV host is synchronously acquired, and the triaxial acceleration signal of the LDV host is acquired using a triaxial accelerometer. The triaxial acceleration signal is integrated to obtain a reference velocity signal. Using the reference velocity signal as input, an adaptive filtering algorithm is used to subtract the vibration crosstalk component from the original measurement signal to obtain the compensated target vibration signal. By using a polarization-spatial dual diversity receiver module, the orthogonal polarization components in the return light are separated by a polarization beam splitter, and the return light from different spatial locations is received through different pixel regions of the array detector to obtain multi-channel interference signals. Calculate the carrier-to-noise ratio of the interference signals in each channel, determine the fusion weights accordingly, and perform weighted fusion to obtain the fused measurement signal; The fused measurement signal is iteratively used as the original measurement signal to form a signal processing closed loop.

[0008] Preferably, the variable stiffness active vibration isolation module is integrated in parallel with a flexible hinge vibration isolation table and a piezoelectric ceramic actuator. When the piezoelectric ceramic actuator is powered on, it locks all the degrees of freedom of the flexible hinge vibration isolation table, and when the power is off, it releases all the degrees of freedom of the table.

[0009] Preferably, the adaptive filtering algorithm is a normalized least mean square algorithm, and its weight update formula is: in, W ( n ) is the first n The filter weight vector at time t. This is the normalization step size factor. α To prevent extremely small positive numbers with a denominator of zero, X ( n () is the delay tap input vector constructed from the reference velocity signal. e ( n ) represents the error signal; The delay tap input vector is generated by the reference velocity signal at the current moment and before. M The values ​​at time -1 constitute the total. M Let the filter order be . M = ceil ( f s / f L ),in f s The sampling rate of the triaxial acceleration signal is [value]. f L The preset lower limit of the vibration frequency, ceil This is the floor function.

[0010] Preferably, in the polarization-spatial dual diversity receiving module, a polarization beam splitter separates the returned light into... P polarization components andS The polarization components are respectively sent to different channels of the array detector, and the receiving optical path of the polarization-space diversity receiving module does not contain a Faraday rotator.

[0011] Preferably, the determination of the fusion weights specifically involves: in, w k For the first k Channel fusion weights CNR k For the first k The carrier-to-noise ratio of the channel, N ch The total number of channels is determined by the product of the number of channels in the polarization diversity and the number of effective pixel regions in the array detector. I k For the first k Interference signals in the channel, I fusion This is the fused measurement signal.

[0012] Preferably, the calculation of the carrier-to-noise ratio includes: performing a Fourier transform on the interference signal of each channel, identifying the target vibration peak in the spectrum, using the peak power as the signal power, using the average power outside the preset bandwidth on both sides of the peak as the noise power, and calculating the ratio of the two as the carrier-to-noise ratio of the channel. The preset bandwidth is determined based on the frequency resolution of the Fourier transform.

[0013] Preferably, when the piezoelectric ceramic actuator is energized, the displacement is amplified by the lever amplification mechanism and then applied to the locking part of the flexible hinge vibration isolation table to lock all its degrees of freedom of motion.

[0014] A laser vibration measurement crosstalk elimination and dynamic compensation system for an inspection robot using the above method includes: a laser Doppler vibration meter host; The variable stiffness active vibration isolation module is located between the inspection robot base and the laser Doppler vibration meter host, and is configured to switch between a rigid locking state and a flexible vibration isolation state according to the motion state signal. A triaxial accelerometer is fixedly mounted on the main unit of the laser Doppler vibration meter and is used to collect the triaxial acceleration signals of the main unit of the laser Doppler vibration meter. A polarization-spatial dual diversity receiving module includes a polarization beam splitter and an array detector. The polarization beam splitter is used to separate the orthogonal polarization components in the return light, and the array detector is used to receive the return light from different spatial locations through different pixel regions to obtain multi-channel interference signals. The signal processing and control unit is configured to acquire the motion state signal of the inspection robot and send a state switching command to the variable stiffness active vibration isolation module; it is also configured to synchronously receive the original measurement signal of the laser Doppler vibration meter host and the triaxial acceleration signal in the initial state. The reference velocity signal is obtained by integrating the triaxial acceleration signal. The normalized minimum mean square adaptive filtering algorithm is used to subtract the vibration crosstalk component from the original measurement signal using the reference velocity signal as input, so as to obtain the compensated target vibration signal. The carrier-to-noise ratio of the interference signals in each channel is calculated, and the fusion weight is determined accordingly. The multi-channel interference signals are then weighted and fused to obtain the fused measurement signal. After obtaining the fused measurement signal, the fused measurement signal is iteratively used as the original measurement signal to form a signal processing closed loop; The calculation of the carrier-to-noise ratio includes: performing a Fourier transform on the interference signal of each channel, identifying the target vibration peak in the spectrum, using the peak power as the signal power, using the average power outside the preset bandwidth on both sides of the peak as the noise power, and calculating the ratio between the two.

[0015] A computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the above-described method for eliminating and dynamically compensating for laser vibration crosstalk in inspection robots.

[0016] An inspection robot includes the aforementioned laser vibration crosstalk elimination and dynamic compensation system for inspection robots, and a robot base for supporting the system, the robot base including a walking mechanism.

[0017] By adopting the above technical solution, the present invention has the following beneficial effects.

[0018] (1) This invention employs a variable stiffness active vibration isolation module and an adaptive filtering scheme to resolve the contradictory requirements of vibration isolation stiffness in both movement and measurement states of airborne laser vibration measurement. When the robot moves, the variable stiffness active vibration isolation module switches to a rigid locking state, and the laser Doppler vibration meter host is fixedly connected to the base to ensure stable movement. When the robot stops and enters the measurement state, the same module switches to a flexible vibration isolation state to provide low stiffness vibration isolation. At the same time, the triaxial accelerometer collects the residual vibration of the laser Doppler vibration meter host itself, and the adaptive filtering uses the integrated reference velocity signal as input to deduct the vibration crosstalk component from the original measurement signal in real time. Active vibration isolation attenuates vibration at the physical layer, and adaptive filtering eliminates residual vibration at the signal layer. The two are connected in series to form a full-band vibration suppression chain, and the measurement accuracy is not limited by platform vibration.

[0019] (2) This invention utilizes a dual architecture of polarization diversity and spatial diversity to suppress speckle noise. The polarization beamsplitter separates orthogonal polarization components from the reflected light, eliminating the need for a Faraday rotator in the optical path. The array detector receives scattered light from different locations on the target surface through different pixel regions, and the speckle patterns in each region are statistically independent. The total number of channels in the system is equal to the product of the number of polarization channels and the number of spatial channels, effectively doubling the number of channels. The interference signals of each channel are weighted and fused according to the carrier-to-noise ratio (CNR), with channels having a higher CNR receiving a greater weight, while low-quality channels are automatically marginalized. This mechanism can still filter out high-quality signals even during speckle fading or local depolarization, directly eliminating the spike pulse interference that frequently occurs during single-channel reception and improving the signal-to-noise ratio of the demodulated signal.

[0020] (3) The adaptive filtering of this invention adopts the normalized least mean square algorithm, and its weight vector continuously approximates the real mechanical transfer function through iterative updates. The filter order is determined by rounding up the ratio of the triaxial acceleration signal sampling rate to the preset vibration frequency lower limit. The delay tap input vector is composed of the values ​​of the reference velocity signal at the current moment and multiple previous moments, effectively capturing the dynamic characteristics of vibration crosstalk. The algorithm responds quickly to changes in signal statistics and maintains stable convergence under time-varying vibration spectrum conditions, ensuring accurate removal of crosstalk components from the original measurement signal. The variable stiffness active vibration isolation module of this invention is composed of a flexible hinge vibration isolation table and a piezoelectric ceramic actuator connected in parallel. When the piezoelectric ceramic is energized, the lever amplification mechanism amplifies the small displacement and applies it to the locking part of the flexible hinge, locking all its motion degrees of freedom. After the piezoelectric ceramic is de-energized, it returns to its original length, and the flexible hinge releases and restores its elastic deformation capability. This structure provides a locking force to overcome the robot's inertia in the rigid locking state, and the stiffness is reduced to the order of 103 N / m in the flexible vibration isolation state, taking into account both mobility reliability and measurement vibration isolation efficiency.

[0021] (4) The polarization-space dual diversity receiving module of the present invention uses only a polarization beam splitter to separate the P-polarization component and the S-polarization component, without the need for complex components such as Faraday rotators. This simple optical path, combined with the spatial diversity of the array detector, realizes the parallel acquisition of a large number of diversity channels within a limited volume. The total number of channels is determined by the product of the number of polarization diversity channels and the number of effective pixel areas of the array detector, enabling the system to achieve outstanding speckle noise suppression capability in a compact structure, and making it easy to integrate on various inspection robot platforms. On this basis, the present invention deeply links the state perception of the vibration isolation module with the adaptive filtering algorithm. The signal processing and control unit obtains the real-time state of the variable stiffness active vibration isolation module (rigid locking or flexible vibration isolation), and dynamically adjusts the parameters of the adaptive filtering algorithm or starts or stops it according to the state. When the system is in the rigid locking state, the adaptive filtering is paused or run at a very low gain to avoid the algorithm from erroneous convergence due to drastic changes in the characteristics of the input signal in the non-measurement state. When switching back to the flexible vibration isolation state, the preset optimization parameters are quickly restored to ensure that the compensation algorithm is put into operation in the best state. This mechanism solves the problem of inconsistent performance and easy misalignment of the compensation algorithm before and after state switching, and significantly improves the robustness and compensation accuracy of the system when repeatedly switching between multiple states of "movement-measurement".

[0022] (5) This invention constructs a collaborative scheme of variable stiffness active vibration isolation and adaptive filtering. The variable stiffness active vibration isolation module dynamically switches between two states: rigid locking and flexible vibration isolation. Combined with the real-time acquisition and adaptive filtering compensation of the vibration of the LDV host itself by the triaxial accelerometer, a full-band vibration crosstalk suppression chain is formed from the two levels of physical attenuation and signal elimination. This fundamentally solves the contradiction between walking stability and measurement vibration isolation accuracy. At the same time, a polarization-space dual diversity receiving architecture is formed by polarization beam splitter and array detector. A large number of diversity channels are acquired in parallel with a simple optical path without optical rotators. Dynamic weighted fusion is performed according to the carrier-to-noise ratio of each channel. The high signal-to-noise ratio signal after fusion is iteratively used as the original input of adaptive filtering, forming a signal processing closed loop from optical noise reduction to electrical compensation. Attached Figure Description

[0023] The following provides a detailed discussion of the manufacture and application of preferred embodiments of the present invention. However, it should be understood that the present invention provides many applicable inventive concepts that can be embodied in various specific environments. The specific embodiments discussed are merely illustrative of specific ways of manufacturing and using the present invention and do not limit the scope of the invention. For those skilled in the art, other drawings can be obtained from these drawings without any creative effort.

[0024] Figure 1 This is a flowchart of the method of the present invention.

[0025] Figure 2This is a system structure diagram provided for an embodiment of the present invention.

[0026] Figure 3 A comparison diagram of vibration signals with and without diversity schemes provided in an embodiment of the present invention.

[0027] Among them, 1-laser; 2-fiber coupler; 3-acoustic-optic modulator; 4-collimator one; 5-half-wave plate one; 6-polarizing beam splitter one; 7-plane mirror; 8-focusing mirror; 9-target object; 10-collimator two; 11-half-wave plate two; 12-polarizing beam splitter two; 13-half-wave plate three; 14-polarizing beam splitter three; 15-photodetector one; 16-photodetector two; 17-half-wave plate four; 18-polarizing beam splitter four; 19-photodetector three; 20-photodetector four. Detailed Implementation

[0028] The following provides a detailed discussion of the manufacture and application of preferred embodiments of the present invention. However, it should be understood that the present invention provides many applicable inventive concepts that can be embodied in various specific environments. The specific embodiments discussed are merely illustrative of specific ways of manufacturing and using the invention and do not limit the scope of the invention.

[0029] This invention provides a method for eliminating and dynamically compensating for laser vibration measurement crosstalk in inspection robots. For example... Figure 1 As shown. The method flow of this embodiment specifically includes the following steps: Step S1, acquiring the motion state signal of the inspection robot.

[0030] The signal processing and control unit receives real-time drive signals from the inspection robot's walking motors or inertial measurement unit as motion status signals. These signals determine whether the inspection robot is currently in a moving or measuring state. A moving state indicates the robot is walking or overcoming obstacles. A measuring state indicates the robot has come to a stop and entered a stable support position.

[0031] Step S2: Control the variable stiffness active vibration isolation module to switch working states according to the motion state signal.

[0032] The variable stiffness active vibration isolation module integrates a flexible hinge isolation table and a piezoelectric ceramic actuator in parallel. The flexible hinge isolation table adopts a straight-circular flexible hinge structure, providing flexible vibration isolation capabilities in the X, Y, and Z directions. The piezoelectric ceramic actuator uses stacked piezoelectric ceramics and is installed in parallel with the flexible hinge isolation table. The transmission mechanism uses a lever amplification mechanism to amplify the minute displacement of the piezoelectric ceramics and apply it to the locking part of the flexible hinge.

[0033] When the motion status signal indicates that the inspection robot is in motion, the signal processing and control unit applies the rated operating voltage to the piezoelectric ceramic actuator. Upon energization, the piezoelectric ceramic extends, locking all degrees of freedom of the flexible hinge via a lever amplification mechanism. At this point, the laser Doppler vibration meter main unit and the robot base form a rigid integrated structure. The rated operating voltage of the piezoelectric ceramic is calculated and determined based on the robot's maximum acceleration during movement and the mass of the laser Doppler vibration meter main unit, ensuring that the locking force is greater than the maximum inertial force. In this embodiment, the operating voltage is 100V.

[0034] When the motion status signal indicates that the inspection robot is in measurement mode, the signal processing and control unit cuts off the power supply to the piezoelectric ceramic actuator. The piezoelectric ceramic returns to its original length, the transmission mechanism releases, and the flexible hinge vibration isolation table regains its free elastic deformation capability. At this time, the vibration isolation table provides low-stiffness, high-damping flexible vibration isolation for the laser Doppler vibration meter host, with stiffness as low as 10. 3 On the order of N / m.

[0035] Step S2A: Based on the state of the variable stiffness active vibration isolation module, pre-configure the adaptive filtering algorithm parameters.

[0036] The signal processing and control unit acquires the current status signal of the variable stiffness active vibration isolation module. When this signal indicates that the system has switched to flexible vibration isolation, the signal processing and control unit applies the normalized step size factor of the normalized least mean square algorithm in subsequent step S4. Set to a first value (e.g., 0.3) and enable the iterative update process of the algorithm to obtain optimal compensation accuracy and dynamic response. When this signal indicates that the current state has been switched to rigid locking, the signal processing and control unit will... Setting it to a second value (e.g., 0) stops the iterative weight update process of the adaptive filtering algorithm, preventing erroneous convergence of the filter due to drastic changes in input signal characteristics during non-measurement states. In another embodiment, when in a rigid-locked state, it can also be set to... Setting the value to a value much smaller than the first value (e.g., 0.001) allows the algorithm to track at a very low speed, providing a smoother transition for the next switch back to the flexible state. This linkage mechanism ensures that the algorithm's operating state always matches the physical boundary conditions of the system.

[0037] Step S3: In the measurement state, acquire the original measurement signal and triaxial acceleration signal of the laser Doppler vibration meter host.

[0038] The triaxial accelerometer uses a MEMS capacitive triaxial accelerometer with a range of ±10g, a bandwidth of DC to 1kHz, and a sampling rate of no less than 10kHz. This sensor is fixed to the housing of the laser Doppler vibration meter and acquires vibration acceleration signals in the X, Y, and Z directions of the meter in real time. The acquired acceleration signals are converted into digital signals by the built-in amplification circuit and analog-to-digital converter in the signal processing and control unit.

[0039] Simultaneously, the raw measurement signal from the laser Doppler vibration meter main unit is also sent to the signal processing and control unit after analog-to-digital conversion. The raw measurement signal is the target vibration velocity signal output by the laser Doppler vibration meter main unit, which includes vibration crosstalk components.

[0040] In the initial state, when the system has just entered the measurement state and has not yet obtained the fused measurement signal, the original measurement signal is the signal directly output by the laser Doppler vibrometer host without diversity fusion processing. After obtaining the fused measurement signal, the original measurement signal will be replaced by the fused measurement signal.

[0041] Step S4: Integrate the triaxial acceleration signal to obtain the reference velocity signal, and use adaptive filtering to subtract the vibration crosstalk component.

[0042] The signal processing and control unit performs DC-free processing on the triaxial acceleration digital signals. Then, it uses the trapezoidal numerical integration method to integrate the signals and obtain the time-domain vibration velocity signals of the laser Doppler vibration meter host in the X, Y, and Z directions, which are used as reference velocity signals.

[0043] The adaptive filtering uses the normalized least mean square algorithm and is controlled by the parameters set in step S2A. The specific implementation is as follows.

[0044] Step S4-1: Synchronously acquire the raw velocity signal of the laser Doppler vibration meter host. d ( n and triaxial acceleration a x ( n ), a y ( n ), a z ( n ), n Indicates the first n Each sampling time.

[0045] Step S4-2: Perform DC removal and trapezoidal numerical integration on the triaxial acceleration to obtain the triaxial velocity signal. v i ( n ),in i = x ,y , z .

[0046] Step S4-3: Construct the delayed tap input vector X ( n and weight vector W ( n ). X ( n The reference velocity signal at the current moment and before M The values ​​at time -1 constitute the total value. X ( n Specifically: X ( n )=[ v x ( n ), v x ( n -1),..., v x ( n - M +1), v y ( n ), v y ( n -1),..., v y ( n - M +1), v z ( n ), v z ( n -1),..., v z ( n - M +1)] T M Let the filter order be . M = ceil ( f s / f L ).in f s The sampling rate is the three-axis acceleration signal. f L This is the preset lower limit of the vibration frequency. ceil This is the floor function. In this embodiment... f s =10kHz, fL =200Hz, calculated M =50. Initial time W (0) = 0.

[0047] Step S4-4: Calculate the output of the adaptive filter. y ( n ), which is the crosstalk estimate at the current moment. y ( n )= W T ( n ) X ( n ) Step S4-5: Calculate the error signal e ( n The output is the compensated target vibration signal. e ( n )= d ( n )- y ( n ).

[0048] Steps S4-6: Update the weight vector using the NLMS algorithm to approximate the real mechanical transfer function.

[0049] in, W ( n ) is the first n The filter weight vector at time step; This is the normalization step size factor; its specific value is dynamically set by step S2A based on the current state of the variable stiffness active vibration isolation module. X ( n ) is the delayed tap input vector constructed from the reference velocity signal; e ( n ) represents the error signal; α To prevent extremely small positive numbers with a denominator of zero, this embodiment uses 10. -6 。 | X ( n )| 2 = X ( n ) T X ( n ) represents the instantaneous power of the reference signal.

[0050] Step S4-7: Determine the convergence condition. If the change in the weight vector within 200 consecutive sampling points is less than the convergence threshold... ε , that is | W ( n+1)- W ( n )|< ε Then output e ( n () as the compensated target vibration signal. ε For a very small positive number, we take 10 in this embodiment. -6 Otherwise, return to step S4-4 and continue iterating.

[0051] In another embodiment, adaptive filtering employs a recursive least squares algorithm, using the reference velocity signal as the input, to subtract the vibration crosstalk component from the original measurement signal. The recursive least squares algorithm updates the filter coefficients recursively, using the minimum weighted sum of squared errors as the criterion, resulting in faster convergence for non-stationary signals. In yet another embodiment, adaptive filtering employs a Kalman filter algorithm, using the reference velocity signal as the observation value to establish a state-space model of the vibration crosstalk, and dynamically estimating and compensating for it through a prediction-update loop.

[0052] Step S5: Use the polarization-space dual diversity receiver module to obtain multi-channel interference signals.

[0053] Figure 2 The diagram below illustrates the system structure of an embodiment of the present invention. In this embodiment, the working principle and optical path structure of the laser heterodyne interferometry polarization diversity system are as follows.

[0054] Laser 1 emits a polarized beam, which is split into a reference beam and a measurement beam by fiber coupler 2. An acousto-optic modulator 3 is installed in the reference arm to perform frequency modulation of the reference beam at a fixed fundamental frequency, thereby achieving the frequency shift required for heterodyne interference.

[0055] Measurement optical path: After collimation by collimator 4, the measurement light is adjusted to a 45° linear polarization state by half-wave plate 5. A polarization beam splitter 6, a plane mirror 7, and a focusing mirror 8 are sequentially arranged along the optical path. The measurement beam is split into a transmitted P-polarized component and a reflected S-polarized component by polarization beam splitter 6. The S-polarized component is deflected by 90° by plane mirror 7 and, together with the transmitted P-polarized component, is focused onto the surface of target object 9 by focusing mirror 8. The reflected light scattered from the surface of target object 9 is collected again by focusing mirror 8. The reflected light with different polarization states is combined at polarization beam splitter 6.

[0056] Reference optical path: After being collimated by collimator 210, the reference light is adjusted to a 45° linear polarization state by half-wave plate 211, and then enters polarization beam splitter 212.

[0057] Beam combining and polarization diversity reception: The return light output from the beam combiner 6 contains P-polarization components and S-polarization components. This return light enters the polarization beam combiner 12 and is combined with the reference light at the polarization beam combiner 12 according to the polarization state, and then split into two polarization channels: the P-polarization channel and the S-polarization channel.

[0058] Balanced detection in the P-polarization channel: The combined beam from the P-polarization channel first enters a half-wave plate (13), whose fast axis makes a 22.5° angle with the optical axis. This angle is used to rotate the polarization direction to achieve the splitting ratio required for balanced detection. The rotated polarized beam then enters a polarization beam splitter (14), which splits the beam into two equal-power output beams according to their polarization states. These beams are then fed into photodetector 15 and photodetector 2 (16), respectively. The optical power output of photodetector 15 is denoted as... I 11 The optical power output of photodetector 216 is denoted as I 12 Differential processing is performed on the two outputs, i.e. I 11 minus I 12 The differential interference signal of the P-polarization channel is obtained. This differential processing can effectively suppress common-mode noise and improve the signal-to-noise ratio.

[0059] Balanced detection in the S-polarization channel: The combined beam from the S-polarization channel first enters the half-wave plate 417, whose fast axis also forms a 22.5° angle with the optical axis. This angle is used to rotate the polarization direction to achieve the splitting ratio required for balanced detection. The rotated polarized beam then enters the polarization beam splitter 418, which splits the beam into two equal-power output beams according to their polarization states. These beams are then fed into photodetectors 319 and 420, respectively. The optical power output of photodetector 319 is denoted as... I 21 The optical power output of photodetector 420 is denoted as I 22 Differential processing is performed on the two outputs, i.e. I 21 minus I 22 The differential interference signal of the S-polarization channel is obtained. This differential processing can also effectively suppress common-mode noise and improve the signal-to-noise ratio.

[0060] The aforementioned polarization diversity optical path only uses polarization beamsplitter 1-6, polarization beamsplitter 2-12, polarization beamsplitter 3-14, polarization beamsplitter 4-18, and half-wave plates 1-5, 2-11, 3-13, and 4-17; it does not include a Faraday rotator. To further clarify the physical mechanism of polarization interferometric detection, the Jones matrix is ​​used to quantitatively describe the system's optical path. Let the Jones vector of the polarized light emitted from laser 1 be... Ein Its expression is: in, A The complex amplitude of the light vector. a Let be the amplitude and be the initial phase. The Jones matrices of the polarization state of the polarization beam splitter for transmission and reflection are respectively expressed as: and .

[0061] The reflected light from the surface of target object 9 contains both P-polarization and S-polarization components. After being combined by polarization beam splitter 6, the Jones vector of the reflected light is measured. E m It can be represented as: in, A mp To measure the complex amplitude of the P-polarized component in light, A ms To measure the complex amplitude of the S-polarization component in light, a mp and a ms These are the corresponding amplitudes. To measure the phase of light.

[0062] After the reference light is transmitted through the optical path, its Jones vector E r It can be written as: in, a r The amplitude of the reference light; The phase of the reference light; I It is the imaginary unit.

[0063] The reference light and the measured return light are split into two channels according to their polarization states at polarization beam splitter 12. Taking the channel where photodetector 15 is located as an example, the light vector after beam combining is... E comb1 It consists of the P-polarization component in the measured return light and the S-polarization component in the reference light, i.e. ,in A rs The component of the reference light in the S-polarization direction; A mp To measure the complex amplitude of the P-polarized component in the light, the combined beam passes through the fast axis and... x The Jones matrix of a half-wave plate 313 with an axial angle of 22.5°. J HW for:

[0064] The light vector received by photodetector-15 E 11 for:

[0065] Similarly, the output optical powers of the four photodetectors—photodetector 15, photodetector 26, photodetector 39, and photodetector 40—correspond to the interference signals of the P-polarization and S-polarization channels, respectively. Specifically, the output optical powers of photodetectors 15 and 26 in the P-polarization channel... I 11 and I 12 The output optical power of photodetector 319 and photodetector 420 in the S-polarization channel I 21 and I 22 They are respectively: in, a mp and a ms These measures the amplitudes of the P-polarized and S-polarized components of light, respectively. a rp and a rs These are the amplitudes of the P-polarization component and the S-polarization component in the reference light, respectively. For the phase of the reference light, To measure the phase of light.

[0066] The interference signals from the two orthogonal polarization channels are subjected to differential processing, i.e. I 11 minus I 12 The differential interference signal of the P-polarization channel is obtained. I 21 minus I 22 The differential interference signal of the S-polarization channel is obtained. Differential processing can effectively suppress common-mode noise and improve the signal-to-noise ratio. The above polarization diversity optical path only uses a polarization beam splitter and a half-wave plate, and does not include a Faraday rotator.

[0067] Spatial diversity is further introduced based on polarization diversity. The photodetectors used—photodetector 15, photodetector 26, photodetector 39, and photodetector 420—are all array detectors, employing a planar CMOS array with a resolution of [resolution missing]. M Line × NColumn pixels. Each pixel region corresponds to a slightly different tiny region on the surface of the target object 9. Due to their different spatial locations, each region produces a statistically independent speckle pattern. The array detectors acquire interference signals from all channels in parallel. After polarization diversity and spatial diversity are cascaded, the total number of channels in the system is... N ch =2 M × N In this embodiment, the CMOS resolution is 512×512 pixels, with each 32×32 pixel area merged into one effective pixel region. The spatial diversity channel count is 256, and the total system channel count is... N ch =512.

[0068] Step S6: Calculate the carrier-to-noise ratio of the interference signals in each channel and perform weighted fusion.

[0069] After receiving the interference signals from each channel of the array detector, the signal processing and control unit calculates the carrier-to-noise ratio (CNR) for each channel. The CNR is calculated as follows: for the first channel... k The interference signal from the channel is subjected to a Fourier transform to obtain its spectrum. The Doppler shift spectral peak caused by the target vibration is identified in the spectrum, and the power of this peak is taken as the signal power. The average power outside a predetermined bandwidth on both sides of the peak is taken as the noise power. The ratio of signal power to noise power is calculated as the carrier-to-noise ratio of the channel. CNR k .

[0070] The preset bandwidth value is the target vibration frequency resolution. It is an integer multiple of Δf. In this embodiment, 3Δf is used. Δf is determined by the frequency resolution of the Fourier transform. = f s / N ,in f s The sampling rate of the triaxial acceleration signal. N This represents the number of points in the Fourier transform. In this embodiment... N =2048, f s =10kHz, ≈4.88Hz, preset bandwidth≈14.65Hz.

[0071] After obtaining the carrier-to-noise ratio of each channel, the fusion weight of each channel is calculated. k Channel fusion weight w k : in, CNR k For the first k The carrier-to-noise ratio of the channel; Nch This represents the total number of channels. Channels with a carrier-to-noise ratio (CNR) below a preset threshold are assigned zero weight. The preset threshold is set to 50% of the average CNR of all channels.

[0072] Weighted fusion of multi-channel interferometric signals is performed. The fused measurement signal is then obtained. I fusion : in, w k For the first k Channel fusion weights; I k For the first k Interference signals in the channel; N ch This represents the total number of channels.

[0073] Step S7: Iterate the fused measurement signal as the original measurement signal.

[0074] After obtaining the fused measurement signal, the signal processing and control unit uses this fused measurement signal as the new original measurement signal, replacing the original signal directly output by the LDV host in step S3. In the next adaptive filtering iteration, step S4 will use this fused measurement signal as... d ( n Then, re-execute the vibration crosstalk deduction process.

[0075] Thus, a signal processing closed loop is formed between the high signal-to-noise ratio fused signal output from the polarization-space dual diversity receiving module and the adaptive filtering compensation stage. Dual diversity suppresses speckle noise, and adaptive filtering eliminates vibration crosstalk.

[0076] Figure 3 This is a comparison chart of vibration signals provided in an embodiment of the present invention. The upper curve represents the demodulation result of the velocity signal in single-channel receiving mode, showing frequent spike pulse interference. The lower curve represents the demodulation result of the velocity signal under the dual diversity scheme of the present invention. After enabling polarization diversity and spatial diversity and performing carrier-to-noise ratio weighted fusion processing, spike pulse interference is effectively suppressed.

[0077] The core reason why this embodiment achieves the above-mentioned technical effects lies in the physical-algorithm dual-mode collaborative mechanism between the variable stiffness active vibration isolation module and the adaptive filter. The active vibration isolation module switches the mechanical boundary conditions according to the motion state signal, attenuating most of the vibration energy at the physical layer. The residual low-frequency vibration components are dynamically compensated by the adaptive filter at the electrical layer. The smaller the residual vibration after active vibration isolation attenuation, the faster the convergence speed of the adaptive filter and the lower the steady-state error.

[0078] More importantly, this embodiment introduces a dynamic linkage mechanism between state awareness and algorithm parameters. Through step S2A, the signal processing and control unit forcibly adjusts the core parameters of the adaptive filtering algorithm, namely the step size factor, based on the real-time operating state of the variable stiffness active vibration isolation module. This mechanism solves a hidden but crucial problem in simple physics-algorithm concatenation schemes: when the system frequently switches between movement and measurement states, if the algorithm parameters remain unchanged, abnormal vibration signals acquired in the rigid locking state can easily cause severe imbalance in the filter weight vector. This results in a long reconvergence time for the algorithm after switching back to the measurement state, and may even cause a significant decrease in transient measurement accuracy. By linking states, the algorithm iteration can be paused or slowed down in non-measurement states, fundamentally protecting the accuracy of the filter model. This ensures that the system can start working in the optimal state the instant it enters the measurement state, achieving seamless and high-precision switching between movement and measurement states.

[0079] The polarization-spatial dual diversity receiving module and the adaptive filter work together across modules. The dual diversity module outputs a high signal-to-noise ratio (SNR) measurement signal through carrier-to-noise ratio (CNR) weighted fusion. This signal serves as the raw input to the adaptive filter, reducing the filter's input noise level and improving the accuracy of crosstalk component estimation. The compensated signal output from the adaptive filter can be further fed back to optimize the target vibration spectrum peak identification accuracy in the CNR calculation.

[0080] In a preferred embodiment, the filter order M The order is set between 32 and 128 based on the actual vibration spectrum characteristics. When the vibration spectrum has a simple composition, a smaller order is selected to reduce the computational load. When the vibration spectrum has a complex composition, a larger order is selected to improve the filtering accuracy.

[0081] In another preferred embodiment, the normalized step size factor Set between 0.1 and 0.5. The larger the value, the faster the filter converges, but the larger the steady-state error. A smaller value results in a smaller steady-state error, but slower convergence. In this embodiment, a value of 0.3 is used to balance convergence speed and steady-state accuracy.

[0082] In another embodiment, the array detector uses a planar CCD. A larger pixel-binding area results in fewer spatial diversity channels and higher signal strength per channel. Conversely, a smaller pixel-binding area results in more spatial diversity channels and stronger speckle suppression.

[0083] In another embodiment, rigid locking employs an electromagnetic locking mechanism instead of the piezoelectric ceramic lever amplification mechanism. The electromagnetic locking achieves flexible hinge locking by energizing an electromagnet to attract the locking block. Alternatively, a hydraulic locking mechanism can be used, where a hydraulic cylinder pushes a locking pin to achieve locking.

[0084] In another embodiment, the technical solution of the present invention can be implemented in a complete system. This system includes a Laser Doppler Vibration Meter (LDV) host. The LDV host is responsible for emitting a laser beam towards the target object 9 and receiving the reflected light, outputting a raw measurement signal containing vibration crosstalk components. The LDV host integrates optical components such as a laser 1, fiber optic coupler 2, acousto-optic modulator 3, collimator 1 4, collimator 2 10, half-wave plate 1 5, half-wave plate 2 11, polarization beam splitter 6, plane mirror 7, and focusing mirror 8, as well as all the optical and detection components of the polarization-space dual diversity receiving module.

[0085] The system also includes a variable stiffness active vibration isolation module. This module is positioned between the inspection robot base and the LDV main unit. It consists of a flexible hinge isolation platform and a piezoelectric ceramic actuator connected in parallel. The flexible hinge isolation platform uses a straight-circular flexible hinge structure. The piezoelectric ceramic actuator uses stacked piezoelectric ceramics. The transmission mechanism employs a lever amplification mechanism.

[0086] The variable stiffness active vibration isolation module controls the power supply to and from the piezoelectric ceramic actuator via commands output from the signal processing and control unit. When powered on, the piezoelectric ceramic extends, locking all degrees of freedom of the flexible hinge through a lever amplification mechanism, entering a rigid locking state. When powered off, the piezoelectric ceramic recovers, the flexible hinge releases, and the system enters a flexible vibration isolation state. The system also includes a triaxial accelerometer. The triaxial accelerometer is threadedly mounted on the LDV main unit housing, with the measurement direction aligned with the X, Y, and Z axes of the LDV main unit. This sensor is used to acquire the triaxial acceleration signals of the LDV main unit in real time. The triaxial accelerometer is a MEMS capacitive triaxial accelerometer with a range of ±10g, a bandwidth of DC to 1kHz, and a sampling rate of no less than 10kHz. The system also includes a polarization-spatial dual diversity receiver module. This module is located in the receiving optical path of the LDV main unit and consists of a polarization beam splitter and an array detector.

[0087] A polarization beam splitter is used to separate the reflected light into two orthogonal polarization components: a P-polarization component and an S-polarization component. Its working principle is as follows: The laser 1 emits polarized light, which is split into reference light and measurement light after being split by the fiber coupler 2. The reference light is frequency-shifted by the acousto-optic modulator 3 and then input to collimator 10, waveplate 11, and polarization beamsplitter 12, respectively. The measurement light is split into P-polarized component and S-polarized component by collimator 4, waveplate 5, and polarization beamsplitter 6, respectively, and focused onto the surface of the target object 9. The light scattered back from the target object 9 is combined and then split into two paths according to polarization state at polarization beamsplitter 12. One path enters polarization beamsplitter 14 after passing through waveplate 13, and is received by photodetector 15 and photodetector 26 after being split by polarization beamsplitter 14. The other path enters polarization beamsplitter 18 after passing through waveplate 17, and is received by photodetector 19 and photodetector 20 after being split by polarization beamsplitter 18. The array detector employs a planar CMOS array, receiving reflected light from different spatial locations on the target surface in parallel across different pixel regions to obtain multi-channel interference signals. The system also includes a signal processing and control unit. This unit is the core control and computation component of the entire system, comprising a microprocessor, an analog-to-digital converter, a digital signal processing module, and a drive circuit module. The signal processing and control unit is electrically connected to the variable stiffness active vibration isolation module, the triaxial accelerometer, the polarization-spatial dual diversity receiver module, and the LDV host.

[0088] The signal processing and control unit is configured to acquire the motion state signal of the inspection robot and send a state switching command to the variable stiffness active vibration isolation module. It is also configured to synchronously receive the raw measurement signal and triaxial acceleration signal from the LDV host in the initial state. Furthermore, it is configured to integrate the triaxial acceleration signal to obtain a reference velocity signal, and using a normalized least mean square adaptive filtering algorithm, subtract the vibration crosstalk component from the raw measurement signal using this reference velocity signal as input to obtain the compensated target vibration signal. The unit is further configured to calculate the carrier-to-noise ratio (CNR) of each channel's interference signal, perform a Fourier transform on each channel's interference signal, identify the target vibration spectral peak, use the peak power as the signal power, and the average power outside the preset bandwidth on both sides of the peak as the noise power, calculating the ratio as the CNR of that channel. Finally, it is configured to determine the fusion weights based on this, perform weighted fusion of the multi-channel interference signals to obtain the fused measurement signal, and use the fused measurement signal as the raw measurement signal to form a signal processing closed loop.

[0089] In another embodiment, the robot base includes a locomotion mechanism. In different application scenarios, the locomotion mechanism can be a wheeled chassis, a tracked chassis, or a multi-legged locomotion mechanism. Wheeled or tracked chassis are suitable for structured scenarios such as flat roads, factory areas, or substations. Multi-legged locomotion mechanisms include bipedal, quadrupedal, or hexapedal bionic robots, suitable for complex terrain in the wild. The robot base can also be a humanoid robot, adapting to special inspection needs that require traversing obstacles.

[0090] In a specific test instance, a rough aluminum reflective surface was used as target object 9 to test the vibration measurement accuracy of the proposed solution and the single-channel receiving solution. The test conditions were: the inspection robot moved at a speed of 0.5 m / s and then stopped for measurement; the vibration frequency of target object 9 was 100 Hz, and the vibration velocity amplitude was 1 mm / s. Due to speckle noise, the peak amplitude of spike pulse interference in the demodulated velocity signal of the single-channel receiving solution could reach ±5 mm / s, with a signal-to-noise ratio of 12 dB. After the dual diversity scheme of the proposed solution was activated, spike pulse interference was effectively suppressed, the peak error of the demodulated velocity signal was controlled within ±0.3 mm / s, and the signal-to-noise ratio was improved to 28 dB. After active vibration isolation and adaptive filtering worked together, the measurement error caused by vibration crosstalk decreased from ±0.8 mm / s before compensation to ±0.05 mm / s after compensation, and the crosstalk suppression ratio reached 24 dB. By incorporating a state linkage mechanism, the initial fluctuation amplitude of the measurement signal is reduced by approximately 40% compared to the solution without the linkage mechanism when the robot switches from movement to measurement. The time required to establish stable high-precision measurement is shortened by more than 50%.

[0091] Unlike existing fixed-stiffness active vibration isolation schemes, this embodiment does not rely on force sensor feedback for vibration compensation. Instead, it actively switches the mechanical boundary conditions of the vibration isolation module through state perception, providing optimal stiffness characteristics in both movement and measurement states. Unlike existing polarization diversity schemes that require Faraday rotators, this embodiment uses only a polarization beam splitter to separate orthogonal polarization components and combines this with the spatial diversity of the array detector to achieve parallel acquisition of a large number of diversity channels with a simplified optical path.

[0092] The above embodiments, using a specific control flow and parameter setting as an example, detail the complete implementation process of the method of the present invention. This method can be embedded in a computer-readable storage medium in the form of a software program and executed by a general-purpose or special-purpose computing device. In one embodiment, the technical solution of the present invention can be presented in the form of a computer-readable storage medium. The storage medium stores computer program instructions. When these computer program instructions are executed by one or more processors, the processors complete all or part of the method steps described in the above embodiments. Specifically, the storage medium can be a non-volatile memory, such as ROM, EPROM, EEPROM, flash memory, or solid-state drive. The storage medium can also be a portable storage medium, such as a USB flash drive, external hard drive, or optical disc. The processor can be a general-purpose central processing unit, microcontroller, digital signal processor, or field-programmable gate array. The signal processing and control unit includes the aforementioned processor and memory. The processor calls the program instructions in the memory to execute the steps of acquiring motion state signals, controlling the switching state of the variable stiffness active vibration isolation module, acquiring and processing acceleration and original measurement signals, performing adaptive filtering compensation and carrier-to-noise ratio weighted fusion, and iteratively using the fused signal as the original measurement signal.

[0093] In another embodiment, the technical solution of the present invention can be implemented using a complete inspection robot system. This inspection robot integrates a vibration crosstalk elimination and dynamic compensation system based on laser vibration measurement during movement. The robot body consists of a robot base and a walking mechanism. The walking mechanism can be a wheeled chassis, a tracked chassis, or a legged walking mechanism, depending on the application scenario. An LDV (Laser-Driven Vibration) host is mounted on the upper part of the robot base. A variable stiffness active vibration isolation module is installed between the LDV host and the robot base. One side of the flexible hinge isolation platform of this module is connected to the LDV host, and the other side is connected to the robot base. A three-axis accelerometer is threadedly fastened to the side housing of the LDV host. Its three signal output cables are connected to the analog-to-digital conversion interface of the signal processing and control unit.

[0094] The optical probe of the polarization-spatial diversity receiver module is fixed at the measurement optical path exit of the LDV host. Its array detector's data output cable is connected to the data acquisition interface of the signal processing and control unit. The signal processing and control unit is installed inside the robot's base or in an electrical cabinet on top. This unit communicates with the driver of the walking motor via a CAN bus or RS485 bus, acquiring the walking motor's drive signal as a motion status signal. This unit connects to the drive power module of the piezoelectric ceramic actuator via an analog voltage output interface, sending state switching commands. After the inspection robot starts its inspection task, it walks to the target measurement point, stops, and enters a stable support state. The signal processing and control unit automatically completes vibration isolation state switching, signal acquisition, crosstalk compensation, speckle suppression, and result output, achieving high-precision automated vibration measurement during movement.

[0095] The above embodiments are merely exemplary embodiments of the present invention and are not intended to limit the present invention. The scope of protection of the present invention is defined by the claims. Those skilled in the art can make various modifications or equivalent substitutions to the present invention within its spirit and scope, and such modifications or equivalent substitutions should also be considered to fall within the scope of protection of the present invention. Although the specification has provided a detailed description, it should be understood that various changes, substitutions, and alterations can be made without departing from the spirit and scope of the present invention as defined by the appended claims. Furthermore, the specific embodiments described are not intended to limit the scope of the present invention. Those skilled in the art can readily understand, based on the present invention, that existing or future-developed processes, machines, manufactures, material compositions, means, methods, or steps can perform substantially the same functions or obtain substantially the same results as the embodiments of the present invention. Therefore, the appended claims are intended to include such processes, machines, manufactures, material compositions, means, methods, or steps within their scope.

Claims

1. A method for eliminating and dynamically compensating for vibration crosstalk in laser vibration measurement during the movement of an inspection robot, characterized in that, include: Acquire motion status signals of the inspection robot; A variable stiffness active vibration isolation module is adopted. According to the motion state signal, it switches to a rigid locking state in the moving state, so that the LDV host and the robot base form a rigid integrated structure. In the measurement state, it switches to a flexible vibration isolation state. In the measurement state, the initial raw measurement signal of the LDV host is synchronously acquired, and the triaxial acceleration signal of the LDV host is acquired using a triaxial accelerometer. The triaxial acceleration signal is integrated to obtain a reference velocity signal. Using the reference velocity signal as input, an adaptive filtering algorithm is used to subtract the vibration crosstalk component from the original measurement signal to obtain the compensated target vibration signal. The adaptive filtering algorithm is the normalized least mean square algorithm. By using a polarization-spatial dual diversity receiver module, the orthogonal polarization components in the return light are separated by a polarization beam splitter, and the return light from different spatial locations is received through different pixel regions of the array detector to obtain multi-channel interference signals. Calculate the carrier-to-noise ratio of the interference signals in each channel, determine the fusion weights accordingly, and perform weighted fusion to obtain the fused measurement signal; The fused measurement signal is iteratively used as the original measurement signal to form a signal processing closed loop; The signal processing and control unit acquires the current status signal of the variable stiffness active vibration isolation module; when the current status signal indicates that the variable stiffness active vibration isolation module is in a flexible vibration isolation state, the normalization step size factor of the normalized least mean square algorithm is set to a first value and iterative updates are enabled; when the current status signal indicates that the variable stiffness active vibration isolation module is in a rigid locking state, the normalization step size factor is set to a second value to pause the weight iterative updates. In the polarization-space dual diversity receiving module, the return light is separated into two parts by a polarization beam splitter. P polarization components and S The polarization components are respectively sent to different channels of the array detector, and the receiving optical path of the polarization-space dual diversity receiving module does not contain a Faraday rotator. The calculation of the carrier-to-noise ratio includes: performing a Fourier transform on the interference signal of each channel, identifying the target vibration peak in the spectrum, using the peak power as the signal power, using the average power outside the preset bandwidth on both sides of the peak as the noise power, and calculating the ratio of the two as the carrier-to-noise ratio of the channel; the preset bandwidth is determined according to the frequency resolution of the Fourier transform.

2. The method according to claim 1, characterized in that, The variable stiffness active vibration isolation module is integrated in parallel with a flexible hinge vibration isolation table and a piezoelectric ceramic actuator. When the piezoelectric ceramic actuator is powered on, it locks all the degrees of freedom of the flexible hinge vibration isolation table, and when the power is off, it releases all the degrees of freedom of the table.

3. The method according to claim 1, characterized in that, The weight update formula for the normalized least mean square algorithm is as follows: in, W ( n ) is the first n The filter weight vector at time t. This is the normalization step size factor. α To prevent extremely small positive numbers with a denominator of zero, X ( n () is the delay tap input vector constructed from the reference velocity signal. e ( n ) represents the error signal; The delay tap input vector is generated by the reference velocity signal at the current moment and before. M The values ​​at time -1 constitute the total. M Let the filter order be . M = ceil ( f s / f L ),in f s The sampling rate of the triaxial acceleration signal is [value]. f L The preset lower limit of the vibration frequency, ceil This is the floor function.

4. The method according to claim 1, characterized in that, The determination of the fusion weights is as follows: in, w k For the first k Channel fusion weights CNR k For the first k The carrier-to-noise ratio of the channel, N ch The total number of channels. I k For the first k Interference signals in the channel, I fusion This is the fused measurement signal.

5. The method according to claim 2, characterized in that, When the piezoelectric ceramic actuator is powered on, the displacement is amplified by the lever amplification mechanism and then applied to the locking part of the flexible hinge vibration isolation table to lock all its degrees of freedom of motion.

6. A laser vibration measurement crosstalk elimination and dynamic compensation system for an inspection robot using the method described in any one of claims 1 to 5, characterized in that, include: Laser Doppler vibration meter main unit; The variable stiffness active vibration isolation module is located between the inspection robot base and the laser Doppler vibration meter host, and is configured to switch between a rigid locking state and a flexible vibration isolation state according to the motion state signal. A triaxial accelerometer is fixedly mounted on the main unit of the laser Doppler vibration meter and is used to collect the triaxial acceleration signals of the main unit of the laser Doppler vibration meter. A polarization-spatial dual diversity receiving module includes a polarization beam splitter and an array detector. The polarization beam splitter is used to separate the orthogonal polarization components in the return light, and the array detector is used to receive the return light from different spatial locations through different pixel regions to obtain multi-channel interference signals. The signal processing and control unit is configured to acquire the motion state signal of the inspection robot and send a state switching command to the variable stiffness active vibration isolation module; it is also configured to synchronously receive the original measurement signal of the laser Doppler vibration meter host and the triaxial acceleration signal in the initial state. The reference velocity signal is obtained by integrating the triaxial acceleration signal. The normalized minimum mean square adaptive filtering algorithm is used to subtract the vibration crosstalk component from the original measurement signal using the reference velocity signal as input, so as to obtain the compensated target vibration signal. The carrier-to-noise ratio of the interference signals in each channel is calculated, and the fusion weight is determined accordingly. The multi-channel interference signals are then weighted and fused to obtain the fused measurement signal. After obtaining the fused measurement signal, the fused measurement signal is iteratively used as the original measurement signal to form a signal processing closed loop; The calculation of the carrier-to-noise ratio includes: performing a Fourier transform on the interference signal of each channel, identifying the target vibration peak in the spectrum, using the peak power as the signal power, using the average power outside the preset bandwidth on both sides of the peak as the noise power, and calculating the ratio between the two.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1 to 5.

8. An inspection robot, characterized in that, The system includes the system as described in claim 6, and a robot base for supporting the system, the robot base including a walking mechanism.

Images