A method for synchronous acquisition and multi-axis alignment analysis of bond phase pulse signals

Abstract

This invention discloses a method for synchronous acquisition and multi-axis alignment analysis of key phase pulse signals, belonging to the field of mechanical monitoring and diagnostic technology. The method includes: configuring a synchronous acquisition system to acquire conditioned key phase pulse signals and vibration signals, and extracting shaft speed and load; constructing a time drift monitoring model to compensate the key phase pulse signals in real time, ensuring time synchronization accuracy; inputting the compensated signals into a shaft alignment correction model based on dynamics and finite element analysis to calculate radial and axial deviations under dynamic loads; establishing a nonlinear tolerance threshold model using a nonlinear regression model, and dynamically adjusting the allowable alignment deviation threshold based on real-time shaft speed and load; automatically outputting alignment qualification conclusions or correction suggestions through closed-loop comparison of shaft alignment deviation and threshold, and supporting trigger signal re-acquisition to adapt to changes in operating conditions; realizing real-time, adaptive analysis of alignment deviations in multi-axis systems, improving the operational stability and maintenance efficiency of rotating machinery.

Description

Technical Field

[0001] This invention belongs to the field of mechanical monitoring and diagnostic technology, specifically a method for synchronous acquisition and multi-axis alignment analysis of bond phase pulse signals. Background Technology

[0002] In rotating machinery systems, the alignment of the shaft system directly affects the operational stability, reliability, and lifespan of the equipment. Multi-shaft systems are widely used in various industrial fields, such as steam turbine-generator sets, compressors, pumps, and other equipment. Poor shaft alignment can lead to increased equipment vibration, bearing wear, seal leakage, and even equipment failure.

[0003] Traditional shaft alignment detection methods are mostly based on static measurements, which cannot reflect the dynamic changes during equipment operation in real time. Operating equipment is affected by various factors, such as temperature changes, dynamic load fluctuations, and foundation deformation. These factors can cause dynamic deviations in the shaft system, and static alignment detection cannot meet the needs of real-time monitoring and adjustment. In addition, in multi-axis systems, the mutual influence of each axis makes shaft alignment problems more complex, and traditional methods are difficult to achieve accurate analysis and evaluation of the overall alignment status of multi-axis systems.

[0004] Key phase pulse signals serve as important reference signals in rotating machinery monitoring, used to mark the rotational position of shafts and playing a crucial role in shaft alignment analysis and vibration monitoring. However, during actual acquisition, key phase pulse signals are susceptible to interference from various factors, such as shaft vibration, sensor installation position deviation, and electromagnetic interference, leading to inaccurate and unstable acquired signals, which affects the accuracy of subsequent shaft alignment analysis.

[0005] Current methods for shaft alignment analysis in multi-axis systems have shortcomings in considering dynamic load fluctuations, time drift compensation, and real-time adaptive adjustment. They cannot accurately reflect dynamic deviations during operation, nor can they dynamically adjust allowable deviations in real time based on parameters such as shaft speed and load, thus affecting the accurate judgment of equipment operating status and timely maintenance. Therefore, there is an urgent need for a method that can achieve precise synchronous acquisition of key phase pulse signals and effectively perform dynamic analysis of multi-axis alignment. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention proposes a method for synchronous acquisition and multi-axis alignment analysis of bond phase pulse signals. A synchronous acquisition system is configured to acquire conditioned bond phase pulse signals and vibration signals, and to extract shaft speed and load parameters. A time drift monitoring model is constructed to compensate for the bond phase pulse signals in real time, ensuring time synchronization accuracy. The compensated signals are input into a shaft alignment correction model based on dynamics and finite element analysis to calculate radial and axial deviations under dynamic loads. A nonlinear regression model is used to establish a nonlinear tolerance threshold model, which dynamically adjusts the allowable alignment deviation threshold based on real-time shaft speed and load. Through closed-loop comparison of shaft alignment deviation values ​​with the threshold, an alignment qualification conclusion or correction suggestion is automatically output, and signal re-acquisition is supported to adapt to changes in operating conditions. This achieves real-time, adaptive analysis of alignment deviations in multi-axis systems, improving the operational stability and maintenance efficiency of rotating machinery.

[0007] To achieve the above objectives, the present invention provides the following technical solution:

[0008] A method for synchronous acquisition and multi-axis alignment analysis of bond phase pulse signals, comprising:

[0009] S1: Configure a synchronous acquisition system based on the needs of multi-axis alignment analysis; the synchronous acquisition system includes a key phase sensor, a vibration sensor, and a signal conditioning circuit;

[0010] S2: The key phase pulse signal and vibration signal are acquired through the synchronous acquisition system, pre-processed by the signal conditioning circuit, and the conditioned key phase pulse signal and vibration signal are obtained. The shaft speed and load are then extracted.

[0011] S3: Construct a time drift monitoring model based on the time characteristics and interval of the conditioned bond phase pulse signal;

[0012] S4: Use the time drift monitoring model to compensate for the time drift of the conditioned bond phase pulse signal, and determine whether the compensated bond phase pulse signal meets the standard. If it does, input the compensated bond phase pulse signal and vibration signal into the shaft alignment correction model constructed based on dynamic principles and finite element analysis methods to calculate the shaft alignment deviation under dynamic load.

[0013] S5: A nonlinear tolerance threshold model is established using nonlinear regression analysis, with shaft alignment deviation, real-time shaft speed, and load as inputs, and the output is the allowable alignment deviation threshold.

[0014] S6: Compare the axis alignment deviation with the allowable alignment deviation threshold. If the axis alignment deviation is within the allowable alignment deviation threshold, output the alignment qualified conclusion and end the process; otherwise, generate correction suggestions and trigger resampling.

[0015] Specifically, the steps of S2 include:

[0016] S2.1: The synchronous acquisition system simultaneously acquires the key phase pulse signal and vibration signal according to a preset sampling frequency; the sampling frequency is determined based on the maximum rotational speed of the shaft and the highest frequency component of the vibration signal;

[0017] S2.2: After the key phase pulse signal and vibration signal enter the signal conditioning circuit, they are amplified by the signal amplification module; the signal conditioning circuit includes a signal amplification module, a filtering module, and a signal shaping module; the amplification factor is controlled by adjusting the resistor and capacitor parameters in the amplification circuit.

[0018] S2.3: The amplified key phase pulse signal and vibration signal enter the filtering module. After filtering, the key phase pulse signal and vibration signal enter the signal shaping module. The comparator and Schmitt trigger convert the filtered key phase pulse signal and vibration signal into a standard square wave signal to obtain the conditioned key phase pulse signal and vibration signal.

[0019] S2.4: For the conditioned key phase pulse signal, the shaft speed is calculated by measuring the time interval between adjacent pulses; the shaft speed is the ratio of 60 to the time interval between adjacent pulses.

[0020] S2.5: The real-time load is obtained based on the conditioned vibration signal.

[0021] Specifically, the steps of S3 include:

[0022] S3.1: Perform time characteristic analysis on the conditioned key phase pulse signal, extract the rising edge trigger time of each pulse, and generate a rising edge timestamp sequence;

[0023] S3.2: Based on the rising edge timestamp sequence, the adjacent time interval sequence is obtained by calculating the difference between two adjacent rising edge timestamps. If any adjacent time interval exceeds the preset percentage range of the theoretical period, it is determined to be an abnormal pulse and corrected by linear interpolation.

[0024] S3.3: Standardize the corrected adjacent time interval sequence to obtain a standardized time interval sequence;

[0025] S3.4: The sliding window statistical method is used to calculate the coefficient of variation, skewness, and kurtosis of the standardized time interval series, quantifying the dispersion and distribution pattern of the standardized time interval series;

[0026] S3.5: Perform frequency domain analysis on the standardized time interval sequence, extract the amplitude ratio of the fundamental frequency and the second harmonic, identify the axis misalignment characteristics, and obtain the harmonic ratio;

[0027] S3.6: Calculate the autocorrelation function of the standardized time interval series and analyze its autocorrelation characteristics;

[0028] S3.7: Integrate the coefficient of variation, skewness, kurtosis, harmonic ratio, and autocorrelation characteristics to form the drift characteristics;

[0029] S3.8: Load the pre-built autoregressive moving average model, train it with drift features as input, and generate a time drift monitoring model.

[0030] Specifically, S4 describes using a time drift monitoring model to compensate for the time drift of the conditioned bond phase pulse signal and determining whether the compensated bond phase pulse signal meets the standard, including:

[0031] S4.1: The least mean square adaptive filtering algorithm dynamically adjusts the filter coefficients based on the time drift prediction value output by the time drift monitoring model, and performs real-time compensation on the conditioned key phase pulse signal to obtain the compensated key phase pulse signal. During the compensation process, the time characteristic parameters of the compensated key phase pulse signal are calculated. The time characteristic parameters of the compensated key phase pulse signal include rise time, fall time, pulse width, and period.

[0032] S4.2: Set the standard range for the time characteristic parameters; the standard range includes the error range of the rise time and the fluctuation range of the pulse width;

[0033] S4.3: Compare the time characteristic parameters of the compensated key phase pulse signal with the compliance standard;

[0034] If all time characteristic parameters are within the acceptable range, the compensated key phase pulse signal is deemed to meet the standard.

[0035] If any time characteristic parameter is outside the acceptable range, time drift compensation will continue until the preset maximum number of compensations is reached.

[0036] Specifically, after determining that the compensated key phase pulse signal meets the standard, the following steps are performed:

[0037] A1: Obtain the system parameters of each shaft segment, and establish the translational and rotational motion equations of each shaft segment based on Newton's second law and Euler's equations; the system parameters include the geometric dimensions, material properties, bearing stiffness parameters, and bearing damping parameters of each shaft segment;

[0038] A2: Combine the translational and rotational motion equations of all axis segments to form the global dynamic equations of the multi-axis system; the form of the global dynamic equations is: ,in, Represents the mass matrix, Represents the damping matrix. Represents the stiffness matrix. The external excitation force is generated by the bond phase pulse signal and the vibration signal, and X represents the displacement vector. This represents the first derivative of the displacement vector X with respect to time, i.e., the velocity vector. This represents the second derivative of the displacement vector X with respect to time, i.e., the acceleration vector;

[0039] A3: The multi-axis system is discretized using the finite element analysis method, which divides the multi-axis system into N finite element elements. The nodal displacement and stress-strain relationship of each finite element element are determined to obtain the finite element analysis model of the multi-axis system.

[0040] Specifically, after determining that the compensated key phase pulse signal meets the standard, the following steps are also performed:

[0041] A4: The compensated key phase pulse signal is converted into a speed-time curve and used as the rotational excitation source of the multi-axis system. The input to the finite element analysis model of the multi-axis system is then used as the rotational excitation source. The vibration signal is decomposed into frequency domain components through Fourier transform and mapped to the corresponding nodes of the finite element analysis model of the multi-axis system as external excitation force, thus obtaining the finite element model after loading excitation.

[0042] A5: Based on the finite element model after loading excitation, the Runge-Kutta method is used to solve the global dynamic equations to obtain the displacement, velocity, and acceleration of each node, and output the dynamic response of the multi-axis system; the dynamic response of the multi-axis system includes the axis trajectory, bending moment distribution, and bearing reaction force;

[0043] The axis trajectory is determined based on the dynamic displacement curves of each axis segment;

[0044] The bending moment distribution is calculated based on nodal displacements;

[0045] The bearing reaction force is calculated based on boundary conditions and dynamic response.

[0046] Specifically, after determining that the compensated key phase pulse signal meets the standard, the following steps are also performed:

[0047] A6: The nodal displacements are interpolated using the element shape functions of the finite element model after loading excitation to obtain the strain distribution inside the shaft segment, and then the stress field is obtained by combining the material constitutive relation.

[0048] A7: Based on the nodal displacements output by the finite element model after loading excitation, calculate the offset of the centerline of each shaft segment relative to the ideal axis, take the peak value as the radial deviation, and calculate the axial deviation by combining the axial displacement difference and tilt angle of the coupling end face.

[0049] A8: Integrate the radial and axial deviations calculated from the finite element model after loading excitation, and output a dynamic shaft alignment deviation report; the shaft alignment deviation report includes the deviation amount, deviation location, and the trend of change over time.

[0050] Specifically, the steps of S5 include:

[0051] S5.1: Collect experimental data and preprocess the experimental data; the experimental data includes the system operating status and performance indicators under different shaft alignment deviations, real-time shaft speeds and load conditions;

[0052] S5.2: Load the pre-built nonlinear regression model and train the nonlinear regression model using the preprocessed experimental data. By adjusting the parameters of the nonlinear regression model, the error between the predicted value and the actual value of the nonlinear regression model is minimized, and the trained nonlinear regression model is obtained.

[0053] S5.3: Use the trained nonlinear regression model as a nonlinear tolerance threshold model, and input the real-time shaft alignment deviation, real-time shaft speed, and load, and output the allowable alignment deviation threshold.

[0054] Specifically, the steps of S6 include:

[0055] S6.1: Compare the calculated shaft alignment deviation with the allowable alignment deviation threshold output by the nonlinear tolerance threshold model;

[0056] If both the radial and axial deviations of the shaft alignment are less than the corresponding allowable alignment deviation thresholds, then the alignment is deemed acceptable, the conclusion is recorded in the system log, and the entire analysis process ends.

[0057] If the radial or axial deviation of the shaft alignment deviation is greater than the corresponding allowable alignment deviation threshold, then a correction suggestion is generated based on the magnitude and direction of the shaft alignment deviation, combined with the structure and operating characteristics of the multi-axis system.

[0058] S6.2: Display the correction suggestions on the operation interface, and trigger the re-sampling command to synchronously acquire the key phase pulse signal and vibration signal again for re-analysis and judgment.

[0059] Specifically, the generation of the proposed correction includes:

[0060] S6.1.1: Based on the magnitude and direction of shaft alignment deviation, and combined with the structure and operating characteristics of the multi-axis system, establish a correction suggestion knowledge base; the correction suggestion knowledge base contains correction methods and empirical data under different shaft alignment deviation conditions;

[0061] S6.1.2: When the shaft alignment deviation is detected to exceed the allowable alignment deviation threshold, a matching query is performed in the correction suggestion knowledge base according to the magnitude and direction of the shaft alignment deviation to find the most matching correction method;

[0062] S6.1.3: When the correction methods in the correction suggestion knowledge base do not match, correction suggestions are generated based on the changing trend of shaft alignment deviation and the operating status of the system.

[0063] S6.1.4: Organize and optimize the generated correction suggestions and display them on the operation interface.

[0064] Compared with the prior art, the beneficial effects of the present invention are:

[0065] 1. This invention proposes a method for synchronous acquisition of bond phase pulse signals and multi-axis alignment analysis. By constructing a time drift monitoring model to compensate for the bond phase pulse signals in real time, it effectively eliminates measurement errors caused by sensor time asynchrony or signal transmission delay in multi-axis systems, and improves the calculation accuracy of axis alignment deviation under dynamic loads. At the same time, the axis alignment correction model based on dynamic principles and finite element analysis, combined with the compensated high-precision signal, can more realistically simulate the complex working conditions of multi-axis systems, thereby avoiding the risk of misjudgment caused by signal distortion.

[0066] 2. This invention proposes a method for synchronous acquisition of key phase pulse signals and multi-axis alignment analysis. It employs nonlinear regression analysis to dynamically establish a nonlinear tolerance threshold model, automatically adjusting the allowable alignment deviation threshold range based on real-time shaft speed and load changes. This overcomes the shortcomings of traditional fixed thresholds, which cannot adapt to fluctuations in operating conditions, thus improving the system's adaptability. Through closed-loop comparison of shaft alignment deviation and deviation threshold, it can not only accurately output alignment compliance conclusions but also automatically generate correction suggestions and trigger signal re-acquisition when deviations occur, forming a complete closed loop of monitoring-analysis-correction. This reduces the maintenance cost and downtime risk of multi-axis systems, ensuring the long-term stable operation of rotating machinery. Attached Figure Description

[0067] Figure 1 This is a schematic diagram of a method for synchronous acquisition and multi-axis alignment analysis of bond phase pulse signals according to the present invention;

[0068] Figure 2 This is a flowchart illustrating the principle of a method for synchronous acquisition and multi-axis alignment analysis of bond phase pulse signals according to the present invention. Detailed Implementation

[0069] Please see Figure 1 and Figure 2 The present invention provides an embodiment of a method for synchronous acquisition and multi-axis alignment analysis of bond phase pulse signals, comprising the following steps:

[0070] S1: Configure a synchronous acquisition system based on the needs of multi-axis alignment analysis; the synchronous acquisition system includes a key phase sensor, a vibration sensor, and a signal conditioning circuit;

[0071] Furthermore, the configuration requirements for multi-axis alignment analysis include:

[0072] (1) Determine the installation position of the key phase sensor according to the number of shafts, and configure at least one key phase sensor for each rotating shaft;

[0073] (2) The vibration sensor adopts a three-phase accelerometer array, arranged in three dimensions: axial, radial, and tangential.

[0074] (3) The signal conditioning circuit includes an anti-aliasing filter and a 24-bit ADC module;

[0075] (4) The synchronous acquisition system adopts the IEEE 1588 precise time protocol to achieve microsecond-level synchronization, and the sampling rate is set to 256 times the shaft speed fundamental frequency according to the Nyquist theorem.

[0076] Furthermore, the synchronization process of the acquisition system includes:

[0077] (1) Based on the number of shafts, the rotational speed range of the shafts and the vibration characteristics of the multi-axis system, determine the required number and installation position of key phase sensors to ensure that each shaft has at least one key phase sensor that can accurately capture the key phase pulse signal of the shaft, and the installation position of the key phase sensor should avoid vibration interference sources on the shaft.

[0078] (2) Select vibration sensors for different shaft vibration frequencies and amplitude ranges. The vibration sensors should have high sensitivity and low noise characteristics. The installation method should ensure that the vibration signal of the shaft can be accurately measured. For rotating shafts, magnetic or adhesive installation can be used.

[0079] (3) Design a signal conditioning circuit, which includes a signal amplification module, a filtering module and a signal shaping module. The signal amplification module selects the amplification factor according to the intensity of the key phase pulse signal and the vibration signal to ensure the signal amplitude. The filtering module adopts a bandpass filter, and its passband frequency range is set according to the frequency characteristics of the key phase pulse signal and the vibration signal to effectively filter out noise interference. The signal shaping module converts the conditioned signal into a standard square wave signal, which is convenient for subsequent time characteristic analysis and processing.

[0080] It should be noted that the installation position of the key phase sensor is determined according to the number of shafts, with at least one key phase sensor configured for each rotating shaft. This configuration can accurately lock the motion characteristics of each rotating shaft. Different rotating shafts will exhibit unique motion states during operation due to their own structure, load, and other factors. Configuring a key phase sensor for each shaft separately can avoid signal interference between shafts and ensure accurate capture of the key phase pulse signal of each shaft. For example, in multi-axis linked mechanical equipment, such as multiple rotor shafts of a steam turbine, if a single key phase sensor is shared, the pulse signals of each shaft will be mixed together, making it impossible to distinguish the operating state of each shaft. However, by configuring a key phase sensor for each shaft separately, the pulse information of each shaft can be clearly obtained, providing an accurate data foundation for subsequent shaft speed calculation, time drift monitoring, etc. This is something that other methods of sharing sensors cannot achieve.

[0081] S2: The key phase pulse signal and vibration signal are acquired through the synchronous acquisition system, pre-processed by the signal conditioning circuit, and the conditioned key phase pulse signal and vibration signal are obtained. The shaft speed and load are then extracted.

[0082] The shaft speed calculation uses a periodic measurement method, which measures the average value of the interval between five consecutive key phase pulses.

[0083] It's important to explain that the shaft speed calculation employs a periodic measurement method, averaging the intervals of five consecutive key phase pulses. This improves the accuracy and stability of the calculation. The interval of a single key phase pulse may be affected by various instantaneous disturbances, but the average of five consecutive intervals effectively offsets these instantaneous errors. For example, if an external disturbance causes an anomaly in the interval of any key phase pulse at any moment, calculating the shaft speed based solely on that interval would result in a significant error. However, by averaging five intervals, the impact of the anomaly is mitigated, and the calculated shaft speed is closer to the true value. This method of averaging multiple measurements makes the shaft speed parameters more reliable, providing a precise speed reference for subsequent time drift monitoring and shaft alignment analysis—an accuracy that cannot be achieved with a single measurement.

[0084] Furthermore, the preprocessing by the signal conditioning circuit specifically includes:

[0085] (1) Perform Schmitt triggering shaping on the bond phase pulse signal, and set the trigger threshold to 30%-70% of the bond phase pulse signal amplitude;

[0086] (2) A moving average filter is used to reduce noise in the vibration signal, and the window length is set to 10 sampling periods;

[0087] It should also be noted that using a moving average filter to denoise the vibration signal, with a window length set to 10 sampling periods, can effectively filter out high-frequency noise while preserving the signal trend. The moving average filter smooths out rapid fluctuations in the signal by averaging 10 consecutive sampling points; these rapid fluctuations are often noise. For example, if high-frequency electromagnetic noise is mixed into the vibration signal, it is significantly reduced after moving average filtering, while the overall trend of the vibration signal is preserved. This allows subsequent vibration signal analysis to focus more on the true vibration characteristics. This ability to preserve the signal trend while reducing noise is difficult to achieve with a single mean filter.

[0088] (3) Eliminate the phase distortion of the vibration signal by using a zero-phase digital filter. The filter order is set to 8th order Butterworth.

[0089] It should also be noted that setting the filter order to 8th-order Butterworth ensures that the signal's phase characteristics remain unchanged. Phase distortion causes a time shift in the vibration signal, affecting the determination of the vibration's occurrence time. The 8th-order Butterworth filter possesses a flat passband and excellent attenuation characteristics. Combined with a zero-phase filtering algorithm, it can remove noise while preserving the signal's phase. For example, when analyzing the time correlation between the vibration signal and the key phase pulse signal, the zero-phase filtered vibration signal accurately corresponds to its rotational position, while a signal with phase distortion leads to incorrect time correspondence, affecting the accuracy of the analysis results. The time accuracy provided by this zero-phase characteristic is higher than that of filters with phase distortion.

[0090] It should be noted that by acquiring key phase pulse signals and vibration signals through a synchronous acquisition system, the simultaneous acquisition of these two signals can be achieved. The key phase pulse signal reflects the shaft's rotational position and velocity information, while the vibration signal reflects the shaft's vibration state. The synchronous acquisition of both establishes a correspondence between rotational position and vibration state. For example, when analyzing the relationship between shaft vibration and rotation angle, synchronously acquired data can accurately display the shaft's vibration amplitude at any rotation angle, thereby determining whether abnormal vibration exists at a specific location. This is crucial for diagnosing problems such as shaft eccentricity and misalignment. Signals acquired asynchronously cannot undergo this precise correlation analysis.

[0091] S3: Construct a time drift monitoring model based on the time characteristics and interval of the conditioned bond phase pulse signal;

[0092] It needs to be explained that the core of the time drift monitoring model is to capture the periodic signal distortion caused by factors such as wear of the mechanical transmission system, sudden load changes, or thermal deformation by quantifying the time series nonlinear characteristics of the bond phase pulse signal. Essentially, it establishes a deviation mapping relationship between the pulse arrival time and the ideal equally spaced sequence, and models the drift amount through time-domain statistical analysis combined with prediction algorithms. Its underlying logic is divided into three levels:

[0093] Signal characterization layer: Transforms the time interval sequence of the key-phase pulse signal into measurable statistical features, such as mean, variance, and higher-order moments;

[0094] Dynamic modeling layer: Captures the drift trend of non-stationary signals through time series analysis algorithms;

[0095] Validation feedback layer: Utilize historical data to validate the model's sensitivity to actual drift.

[0096] S4: Use the time drift monitoring model to compensate for the time drift of the conditioned bond phase pulse signal, and determine whether the compensated bond phase pulse signal meets the standard. If it does, input the compensated bond phase pulse signal and vibration signal into the shaft alignment correction model constructed based on dynamic principles and finite element analysis methods to calculate the shaft alignment deviation under dynamic load.

[0097] It is important to emphasize that the core of the time drift monitoring model is to solve the time series distortion of the bond phase pulse signal, such as the pulse interval deviation caused by mechanical wear and thermal deformation. Its compensation goal is to ensure the consistency of the time reference of the bond phase pulse signal, which is the key to achieving time synchronization in subsequent multi-axis alignment analysis. On the other hand, the vibration signal, through its amplitude, frequency and phase characteristics, focuses on noise reduction, phase fidelity preservation and signal shaping in the preprocessing of the vibration signal, and therefore does not involve the need for time dimension drift compensation.

[0098] S5: A nonlinear tolerance threshold model is established using nonlinear regression analysis, with shaft alignment deviation, real-time shaft speed, and load as inputs, and the output is the allowable alignment deviation threshold.

[0099] S6: Compare the axis alignment deviation with the allowable alignment deviation threshold. If the axis alignment deviation is within the allowable alignment deviation threshold, output the alignment qualified conclusion and end the process; otherwise, generate correction suggestions and trigger resampling.

[0100] The specific steps of S2 include:

[0101] S2.1: The synchronous acquisition system simultaneously acquires the key phase pulse signal and vibration signal according to a preset sampling frequency; the sampling frequency is determined based on the maximum rotational speed of the shaft and the highest frequency component of the vibration signal;

[0102] S2.2: After the key phase pulse signal and vibration signal enter the signal conditioning circuit, they are amplified by the signal amplification module; the signal conditioning circuit includes a signal amplification module, a filtering module, and a signal shaping module; the amplification factor is controlled by adjusting the resistor and capacitor parameters in the amplification circuit.

[0103] S2.3: The amplified key phase pulse signal and vibration signal enter the filtering module. After filtering, the key phase pulse signal and vibration signal enter the signal shaping module. The filter key phase pulse signal and vibration signal are converted into standard square wave signals by a comparator and a Schmitt trigger to obtain the conditioned key phase pulse signal and vibration signal. The center frequency and bandwidth of the bandpass filter in the filtering module are adjusted according to the frequency characteristics of the actual signal to retain the effective signal to the greatest extent and suppress noise. In this invention, the comparator is an LM311.

[0104] Furthermore, the specific steps in S2.3 include:

[0105] (1) Adjust the gain of the filtered key pulse signal to ensure that its peak-to-peak value is within the range of the comparator input voltage. If the amplitude of the filtered key pulse signal is too low, use a programmable gain amplifier to amplify it in stages; if the amplitude is too high, use a resistor divider network to attenuate it.

[0106] (2) DC bias cancellation is applied to the vibration signal to prevent the subsequent comparator from being unable to process it due to negative voltage;

[0107] (3) Add a low-pass RC filter to the input of the comparator to suppress high-frequency glitches. For example, if the frequency of the key phase pulse signal is 1kHz, the cutoff frequency of the filter is set to 1.5kHz.

[0108] (4) Set the reference voltage of the comparator according to the amplitude characteristics of the key phase pulse signal;

[0109] (5) When the input signal voltage is higher than the reference voltage of the comparator, the comparator outputs a high level; when the input signal voltage is lower than the reference voltage of the comparator, the comparator outputs a low level, generating the original square wave signal.

[0110] (6) Set the upper and lower threshold values ​​of the Schmitt trigger; the upper and lower threshold values ​​are set according to twice the signal noise amplitude;

[0111] (7) The Schmitt trigger shapes the original square wave signal output by the comparator, eliminates jitter and sharpens the rising and falling edges. The Schmitt trigger is a prior art in this field and is not an inventive solution of this application. It will not be described in detail here.

[0112] (8) The shaped square wave signal is output through an optocoupler isolator to isolate ground loop interference and provide electrical protection. At the same time, a 22Ω resistor is connected in series and a capacitor is connected in parallel at the output terminal to suppress high-frequency ringing.

[0113] (9) The vibration signal is processed separately in the signal shaping module, including: after passing through the anti-aliasing filter, it is converted into a digital signal by the ADC and output synchronously with the key phase square wave signal to obtain the conditioned key phase pulse signal and vibration signal.

[0114] It should be noted that Schmitt triggering of the key phase pulse signal, with the trigger threshold set to 30%-70% of the key phase pulse signal amplitude, effectively eliminates noise interference in the signal, making the edges of the key phase pulse signal steeper. Schmitt triggering has hysteresis characteristics; when the signal amplitude fluctuates within the threshold range, it will not cause frequent jumps in the output signal. For example, when the key phase pulse signal fluctuates slightly due to noise, with the amplitude varying between 30% and 70% of the pulse signal amplitude, the output of the Schmitt trigger remains stable, avoiding false triggering caused by noise and ensuring the accuracy of the shaped pulse signal. This provides a reliable signal source for subsequent time interval measurements, a noise immunity advantage that simple comparator shaping cannot achieve.

[0115] S2.4: For the conditioned key phase pulse signal, the shaft speed is calculated by measuring the time interval between adjacent pulses; the shaft speed is the ratio of 60 to the time interval between adjacent pulses.

[0116] S2.5: The real-time load is obtained based on the conditioned vibration signal. The real-time load is a predicted value obtained through prediction.

[0117] Furthermore, the process of obtaining real-time load based on the conditioned vibration signal essentially involves indirect measurement through the mapping relationship between vibration characteristics and load. The core logic is to establish a quantitative correlation model between vibration signal characteristic parameters and actual load values. Specific steps include:

[0118] (1) According to the same time base as the key phase pulse signal, the conditioned vibration signal is divided into fixed windows to ensure that the conditioned vibration signal in each window can reflect the stable load state in that period. At the same time, the vibration signal window is precisely aligned with the time interval of shaft speed calculation by the timestamp of the key phase pulse signal to eliminate the influence of time deviation on subsequent correlation analysis.

[0119] (2) For each vibration signal window, calculate the time-domain characteristics strongly correlated with the load, including the root mean square value of vibration, peak factor and pulse amplitude. The root mean square value of vibration is obtained by the root mean square calculation formula. The peak factor is the ratio of the signal peak value to the root mean square value of vibration. The pulse amplitude is the sum of the amplitudes of the pulse signals in the vibration signal that exceed 3 times the root mean square value of vibration. The root mean square calculation formula is the prior art in this field and is not an inventive solution of this application. It will not be elaborated here.

[0120] (3) Perform Fourier transform on the vibration signal of each window, and extract the frequency domain features after converting to the frequency domain. The frequency domain features refer to the fundamental frequency amplitude, the ratio of the second harmonic to the fundamental frequency amplitude, where the fundamental frequency amplitude is calculated from the key phase pulse signal, that is, the vibration amplitude corresponding to the reciprocal of the time interval between adjacent pulses.

[0121] (4) Calculate the correlation coefficients between the time-domain features and frequency-domain features and the known load value, retain the features whose absolute value of the correlation coefficient is greater than or equal to the preset coefficient threshold, and form the final feature set. The known load value is measured by calibration experiment, the correlation coefficient is Pearson correlation coefficient, the preset coefficient threshold is set to 0.7, and the Pearson correlation coefficient is the prior art in this field and is not an inventive solution of this application, so it will not be described in detail here.

[0122] (5) Load the multiple linear regression model and use the final feature set to train the multiple linear regression model to obtain the trained multiple linear regression model;

[0123] (6) The conditioned vibration signal acquired in real time is processed according to the above window division and preprocessing method, and the root mean square value of vibration, fundamental frequency amplitude, and the ratio of second harmonic to fundamental frequency amplitude are extracted. Combined with the calculated shaft speed, the signal is input into the calibrated multiple linear regression model to obtain the predicted value of real-time load.

[0124] The specific steps of S3 include:

[0125] S3.1: Perform time characteristic analysis on the conditioned key phase pulse signal, extract the rising edge trigger time of each pulse, and generate a rising edge timestamp sequence;

[0126] S3.2: Based on the rising edge timestamp sequence, the adjacent time interval sequence is obtained by calculating the difference between two adjacent rising edge timestamps. If any adjacent time interval exceeds the preset percentage range of the theoretical period, it is determined to be an abnormal pulse and corrected by linear interpolation. The linear interpolation is the prior art in this field and is not an inventive solution of this application, so it will not be described in detail here.

[0127] S3.3: Standardize the corrected adjacent time interval sequence to obtain a standardized time interval sequence;

[0128] S3.4: The sliding window statistical method is used to calculate the coefficient of variation, skewness, and kurtosis of the standardized time interval series, quantifying the dispersion and distribution pattern of the standardized time interval series;

[0129] Furthermore, the specific steps in S3.4 include:

[0130] (1) Determine the window length and sliding step size based on the sampling frequency and the characteristic period of the mechanical system;

[0131] (2) Calculate the mean and standard deviation of the standardized time interval sequence within the window. If there are missing values ​​within the window, use linear interpolation of the valid data before and after to fill them.

[0132] (3) Obtain the coefficient of variation from the percentage form of the ratio of the standard deviation to the mean;

[0133] (4) Calculate the cube of each data point in the standardized time interval sequence as the first power variable;

[0134] (5) Calculate the square of each data point in the standardized time interval sequence as the second power variable;

[0135] (6) Calculate the fourth power of each data point in the standardized time interval sequence as the third power variable;

[0136] (7) Calculate the mean of the first power variable of all data in the standardized time interval series to obtain the first sum variable;

[0137] (8) Calculate the mean of the second power variable of all data in the standardized time interval series to obtain the second sum variable;

[0138] (9) Calculate the mean of the third power variable of all data in the standardized time interval series to obtain the third sum variable;

[0139] (10) Take the square root of the second sum variable and then raise it to the cube to obtain the fourth power variable;

[0140] (11) Calculate the ratio of the first sum variable to the fourth power variable to obtain the skewness of the standardized time interval series;

[0141] If the skewness is greater than zero, it indicates that the standardized time interval sequence is right-skewed;

[0142] If the skewness is less than zero, it indicates that the standardized time interval sequence is left-skewed;

[0143] If the absolute value of the skewness is greater than 1, it indicates that the standardized time interval sequence is significantly deviated from a symmetric distribution.

[0144] (12) Calculate the second sum variable squared to obtain the fifth power variable;

[0145] (13) Calculate the difference between the ratio of the third sum variable and the fifth power variable and 3 to obtain the kurtosis of the standardized time interval sequence, where the size of the kurtosis indicates the distribution state;

[0146] If the kurtosis is greater than zero, it indicates that the distribution is sharper than the normal distribution.

[0147] If the kurtosis is less than zero, it indicates that the distribution is flatter than the normal distribution.

[0148] S3.5: Perform frequency domain analysis on the standardized time interval sequence, extract the amplitude ratio of the fundamental frequency and the second harmonic, identify the axis misalignment characteristics, and obtain the harmonic ratio;

[0149] Furthermore, the specific steps in S3.5 for performing frequency domain analysis on the standardized time interval sequence and extracting the amplitude ratio of the fundamental frequency to the second harmonic include:

[0150] (1) Extract sample segments of length 2 to powers of 2 from the standardized time interval sequence and add a Hamming window. If the length is insufficient during the extraction process, pad with zeros at the end until the next power.

[0151] (2) Determine the sampling rate and frequency resolution; the sampling rate is the reciprocal of the actual average interval of the key phase pulse signal; the frequency resolution is the ratio of the sampling rate to the window length;

[0152] (3) Perform a Fourier transform on the windowed standardized time interval sequence to obtain the complex spectrum;

[0153] (4) Based on the ratio of the complex spectrum to the window length, the single-sided amplitude spectrum and the DC component are obtained;

[0154] (5) Convert the Fourier transform points into actual frequencies and establish a frequency-amplitude correspondence;

[0155] (6) Search for the frequency corresponding to the global maximum value in the single-sided amplitude spectrum, which is the fundamental frequency. If there are multiple peaks, select the frequency with the smallest deviation from the shaft speed.

[0156] (7) Search for local peaks at the frequency corresponding to twice the global maximum value and record their amplitudes. If no local peak is detected, set the amplitude of the local peak to the average amplitude of the neighborhood of that frequency point to obtain the second harmonic. The frequency point refers to the frequency corresponding to twice the global maximum value, that is, twice the frequency of the fundamental frequency.

[0157] (8) Calculate the amplitude ratio of the fundamental frequency to the second harmonic. If the ratio is greater than the preset threshold, an early warning will be generated.

[0158] S3.6: Calculate the autocorrelation function of the standardized time interval sequence and analyze its autocorrelation characteristics. The autocorrelation function is prior art in this field and is not an inventive solution of this application. It will not be described in detail here.

[0159] S3.7: Integrate the coefficient of variation, skewness, kurtosis, harmonic ratio, and autocorrelation characteristics to form the drift characteristics;

[0160] S3.8: Load the pre-built autoregressive moving average model, train it with the drift feature as input, and generate a time drift monitoring model. The autoregressive moving average model is existing technology in this field and is not an inventive solution of this application, so it will not be described in detail here.

[0161] The time drift monitoring model described in S4 is used to compensate for the time drift of the conditioned bond phase pulse signal and to determine whether the compensated bond phase pulse signal meets the standard, including:

[0162] S4.1: The least mean square adaptive filtering algorithm dynamically adjusts the filter coefficients based on the time drift prediction value output by the time drift monitoring model, and performs real-time compensation on the conditioned key phase pulse signal to obtain the compensated key phase pulse signal. During the compensation process, the time characteristic parameters of the compensated key phase pulse signal are calculated. The time characteristic parameters of the compensated key phase pulse signal include rise time, fall time, pulse width, and period. The least mean square adaptive filtering algorithm is prior art in this field and is not an inventive solution of this application, so it will not be described in detail here.

[0163] Furthermore, the initial coefficient settings for the least mean square adaptive filtering algorithm include:

[0164] (1) Based on the frequency characteristics of the key phase pulse signal and the approximate range of time drift, make a preliminary estimate of the filter order and initial coefficients;

[0165] (2) A set of initial coefficients is generated using a random number generation method. The range of values ​​for this set of initial coefficients should be determined based on the type and parameters of the filter.

[0166] (3) Normalize the generated initial coefficients to ensure that the values ​​of the initial coefficients are within a reasonable range, and avoid filter instability due to coefficients that are too large or too small.

[0167] (4) Substitute the normalized initial coefficients into the least mean square adaptive filtering algorithm to perform preliminary filtering calculations, observe the filtering effect, and fine-tune the initial coefficients according to the actual situation.

[0168] It should be noted that the least mean square adaptive filtering algorithm dynamically adjusts the filter coefficients based on the time drift prediction value output by the time drift monitoring model, performing real-time compensation on the conditioned key phase pulse signal. This allows it to quickly adapt to signal changes and achieve precise compensation. Time drift is a dynamic process, and fixed-coefficient filters cannot handle this change. For example, when the shaft undergoes thermal deformation due to increased temperature, causing the time drift to gradually increase, the least mean square adaptive filtering algorithm can adjust the coefficients in real time to maintain optimal compensation, while the compensation effect of a fixed-coefficient filter will gradually deteriorate with drift changes. This dynamic adaptability results in superior compensation accuracy compared to fixed filters.

[0169] The initial coefficient setting process of the least mean square adaptive filtering algorithm ensures the stability of the filter and the initial compensation effect. Based on the signal frequency characteristics and drift range, the order and initial coefficients are estimated. Through random number generation and normalization, the initial coefficients are kept within a reasonable range. After preliminary filtering and fine-tuning, the filter can achieve a good compensation effect in the early stages of startup, avoiding filter instability or excessive compensation errors caused by unreasonable initial coefficients. For example, when the filter starts up, reasonable initial coefficients enable it to converge quickly to the optimal state, while unreasonable initial coefficients may cause the filter to oscillate and fail to function properly. The stability and initial effect brought by this proper initial coefficient setting cannot be achieved by arbitrarily setting the initial coefficients.

[0170] S4.2: Set the compliance standard range for the time characteristic parameters; the compliance standard range includes the error range of the rise time and the fluctuation range of the pulse width, providing a clear basis for judging the quality of the compensated signal. These standards can quantify the quality of the signal, avoiding errors in subjective judgment. Only signals that meet these standards are considered qualified, ensuring the signal quality used in subsequent analysis.

[0171] S4.3: Compare the time characteristic parameters of the compensated key phase pulse signal with the compliance standard;

[0172] If all time characteristic parameters are within the acceptable range, the compensated key phase pulse signal is deemed to meet the standard.

[0173] If any time characteristic parameter is outside the acceptable range, time drift compensation will continue until the preset maximum number of compensations is reached.

[0174] It should be noted that comparing the time characteristic parameters of the compensated key phase pulse signal with the compliant standard, and continuing compensation until the maximum number of compensations is reached, ensures the signal quality used for the final analysis. Through multiple compensations, the signal error can be gradually reduced until the standard is met. For example, if the pulse width fluctuation range is 8% after the first compensation, which does not meet the standard, after a second compensation, the fluctuation range drops to 4%, meeting the standard. This ensures that a high-quality signal is used for subsequent shaft alignment deviation calculations. This strict control over signal quality cannot be achieved by stopping compensation after a single compensation.

[0175] After determining that the compensated key phase pulse signal meets the standard, the following steps are performed:

[0176] A1: Obtain the system parameters of each shaft segment, and establish the translational and rotational motion equations of each shaft segment based on Newton's second law and Euler's equations; the system parameters include the geometric dimensions, material properties, bearing stiffness parameters, and bearing damping parameters of each shaft segment. Newton's second law and Euler's equations are existing technologies in this field and are not the inventive solution of this application, and will not be elaborated here.

[0177] Among them, Newton's second law describes the relationship between translational motion and force, while Euler's equations describe the relationship between rotational motion and torque. By combining parameters such as the geometric dimensions and material properties of the shaft segment, the established motion equations can realistically simulate the motion state of the shaft segment under stress.

[0178] Furthermore, based on the mechanical drawings or CAD models of the multi-axis system, determine the length, diameter, material properties, and the location of couplings or bearings for each axis. Material properties include elastic modulus, Poisson's ratio, and density.

[0179] It should be noted that, based on Newton's second law and Euler's equations, the translational or rotational motion equations for each shaft segment are established. Combined with finite element discretization, this allows for a complete simulation of the elastic deformation, bearing stiffness, and damping characteristics of a multi-axis system. In the turbine-generator shaft system, traditional simplified models underestimate the alignment deviation caused by shaft bending. However, this design, through finite element analysis, reduces the calculation errors of radial and axial deviations, providing a precise quantitative basis for alignment adjustment.

[0180] A2: Combine the translational and rotational motion equations of all axis segments to form the global dynamic equations of the multi-axis system; the form of the global dynamic equations is: ,in, Represents the mass matrix, Represents the damping matrix. Represents the stiffness matrix. The external excitation force is generated by the bond phase pulse signal and the vibration signal, and X represents the displacement vector. This represents the first derivative of the displacement vector X with respect to time, i.e., the velocity vector. This represents the second derivative of the displacement vector X with respect to time, i.e., the acceleration vector;

[0181] A3: The multi-axis system is discretized using the finite element analysis method, which divides the multi-axis system into N finite element elements. The nodal displacements and stress-strain relationships of each finite element element are determined to obtain the finite element analysis model of the multi-axis system. The finite element analysis method is existing technology in this field and is not an inventive solution of this application, so it will not be described in detail here.

[0182] A4: The compensated key phase pulse signal is converted into a speed-time curve and used as the rotational excitation source of the multi-axis system. The input to the finite element analysis model of the multi-axis system is then used as the rotational excitation source. The vibration signal is decomposed into frequency domain components through Fourier transform and mapped to the corresponding nodes of the finite element analysis model of the multi-axis system as external excitation force, thus obtaining the finite element model after loading excitation. The Fourier transform is a prior art in this field and is not an inventive solution of this application, so it will not be described in detail here.

[0183] It should be noted that by converting the compensated key phase pulse signal into a speed-time curve and decomposing the vibration signal into frequency domain components through Fourier transform, the alignment deviation changes under unit start-up, shutdown, and sudden load changes can be simulated. During gas turbine startup, traditional static analysis can only obtain steady-state alignment deviation, while this design can capture the dynamic deviation peak caused by thermal deformation during startup, providing crucial data to avoid excessive vibration during the startup phase and reducing the risk of equipment damage during startup.

[0184] A5: Based on the finite element model after loading excitation, the Runge-Kutta method is used to solve the global dynamic equations to obtain the displacement, velocity, and acceleration of each node, and output the dynamic response of the multi-axis system; the dynamic response of the multi-axis system includes the axis trajectory, bending moment distribution, and bearing reaction force. The Runge-Kutta method is existing technology in this field and is not an inventive solution of this application, so it will not be described in detail here.

[0185] The axis trajectory is determined based on the dynamic displacement curves of each axis segment;

[0186] The bending moment distribution is calculated based on nodal displacements;

[0187] The bearing reaction force is calculated based on boundary conditions and dynamic response.

[0188] A6: The nodal displacements are interpolated using the element shape functions of the finite element model after loading excitation to obtain the strain distribution inside the shaft segment, and then the stress field is obtained by combining the material constitutive relation.

[0189] A7: Based on the nodal displacements output by the finite element model after loading excitation, calculate the offset of the centerline of each shaft segment relative to the ideal axis, take the peak value as the radial deviation, and calculate the axial deviation by combining the axial displacement difference and tilt angle of the coupling end face.

[0190] A8: Integrate the radial and axial deviations calculated from the finite element model after loading excitation, and output a dynamic shaft alignment deviation report; the shaft alignment deviation report includes the deviation amount, deviation location, and the trend of change over time.

[0191] The specific steps of S5 include:

[0192] S5.1: Collect experimental data and preprocess the experimental data; the experimental data includes the system operating status and performance indicators under different shaft alignment deviations, real-time shaft speeds and load conditions;

[0193] S5.2: Load the pre-built nonlinear regression model and train the nonlinear regression model using the preprocessed experimental data. By adjusting the parameters of the nonlinear regression model, the error between the predicted value and the actual value of the nonlinear regression model is minimized, and a trained nonlinear regression model is obtained. The nonlinear regression model is the prior art in this field and is not an inventive solution of this application, so it will not be described in detail here.

[0194] Furthermore, the optimization of model parameters in the nonlinear regression analysis includes:

[0195] (1) The parameters of the nonlinear regression model are optimized by using a genetic algorithm. The genetic algorithm is a global optimization algorithm based on the principle of biological evolution, which can effectively avoid getting trapped in local optima.

[0196] (2) Initialize the parameters of the genetic algorithm, including population size, crossover probability, mutation probability, etc. The population size should be determined according to the complexity of the problem and the computing resources. The values ​​of crossover probability and mutation probability are generally between 0.1 and 0.9.

[0197] (3) Generate an initial population and randomly generate multiple sets of model parameters as initial individuals. Each individual in the population represents a set of model parameters.

[0198] (4) Calculate the fitness value of each individual. The fitness value is calculated based on the prediction error of the nonlinear regression model. The smaller the prediction error, the higher the fitness value.

[0199] (5) Select, crossover, and mutate individuals in the population based on their fitness values ​​to generate a new generation of population;

[0200] (6) Repeat the above steps until the termination condition is met, such as reaching the maximum number of iterations or the fitness value no longer significantly improving. The optimal individual obtained at this time is the optimized model parameter.

[0201] S5.3: Use the trained nonlinear regression model as a nonlinear tolerance threshold model, and input the real-time shaft alignment deviation, real-time shaft speed, and load, and output the allowable alignment deviation threshold.

[0202] The specific steps of S6 include:

[0203] S6.1: Compare the calculated shaft alignment deviation with the allowable alignment deviation threshold output by the nonlinear tolerance threshold model;

[0204] If both the radial and axial deviations of the shaft alignment are less than the corresponding allowable alignment deviation thresholds, then an alignment qualified conclusion will be output and recorded in the system log, and the entire analysis process will end.

[0205] If the radial or axial deviation of the shaft alignment deviation is greater than the corresponding allowable alignment deviation threshold, then a correction suggestion is generated based on the magnitude and direction of the shaft alignment deviation, combined with the structure and operating characteristics of the multi-axis system.

[0206] S6.2: Display the correction suggestions on the operation interface, and trigger the re-sampling command to synchronously acquire the key phase pulse signal and vibration signal again for re-analysis and judgment.

[0207] The generation of the proposed corrections includes:

[0208] S6.1.1: Based on the magnitude and direction of shaft alignment deviation, and combined with the structure and operating characteristics of the multi-axis system, establish a correction suggestion knowledge base; the correction suggestion knowledge base contains correction methods and empirical data under different shaft alignment deviation conditions;

[0209] S6.1.2: When the shaft alignment deviation is detected to exceed the allowable alignment deviation threshold, a matching query is performed in the correction suggestion knowledge base according to the magnitude and direction of the shaft alignment deviation to find the most matching correction method;

[0210] S6.1.3: When the correction methods in the correction suggestion knowledge base do not match, correction suggestions are generated based on the changing trend of shaft alignment deviation and the operating status of the system.

[0211] S6.1.4: Organize and optimize the generated correction suggestions and display them on the operation interface.

[0212] The embodiments of the present invention have been described above with reference to the accompanying drawings. However, the present invention is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments under the guidance of the present invention without departing from the spirit and scope of the present invention. All of these variations are within the protection scope of the present invention.

Claims

1. A method for key phase pulse signal synchronous acquisition and multi-axis alignment analysis, characterized in that, The method comprises the following steps: S1: configuring a synchronous acquisition system based on multi-axis alignment analysis requirements; The synchronous acquisition system comprises a key phase sensor, a vibration sensor, and a signal conditioning circuit; S2: acquiring key phase pulse signals and vibration signals through the synchronous acquisition system, pre-processing the key phase pulse signals and the vibration signals through the signal conditioning circuit, obtaining conditioned key phase pulse signals and vibration signals, and extracting shaft speed and load; S3: constructing a time drift monitoring model based on the time characteristics and intervals of the conditioned key phase pulse signals; S4: compensating the conditioned key phase pulse signals for time drift by using the time drift monitoring model, and determining whether the compensated key phase pulse signals meet the standards, if the compensated key phase pulse signals meet the standards, inputting the compensated key phase pulse signals and the vibration signals into an axis alignment correction model constructed based on the kinetic principle and the finite element analysis method, and calculating the axis alignment deviation under dynamic load; S5: establishing a non-linear tolerance threshold model by using non-linear regression analysis, taking the axis alignment deviation, real-time shaft speed, and load as inputs, and outputting the allowable alignment deviation threshold; S6: comparing the axis alignment deviation with the allowable alignment deviation threshold, if the axis alignment deviation is within the allowable alignment deviation threshold, outputting an alignment qualified conclusion and ending the process; Otherwise, generating a correction suggestion and triggering resampling; In S4, the time drift compensation of the conditioned key phase pulse signals by using the time drift monitoring model and the determination of whether the compensated key phase pulse signals meet the standards comprise the following steps: S4.1: a least mean square adaptive filtering algorithm dynamically adjusts the coefficients of the filter according to the time drift prediction value output by the time drift monitoring model, compensates the conditioned key phase pulse signals in real time, obtains the compensated key phase pulse signals, and calculates the time characteristic parameters of the compensated key phase pulse signals in the compensation process; the time characteristic parameters of the compensated key phase pulse signals include rising edge time, falling edge time, pulse width, and period; S4.2: setting a standard range of the time characteristic parameters; the standard range includes an error range of the rising edge time and a fluctuation range of the pulse width; S4.3: comparing the time characteristic parameters of the compensated key phase pulse signals with the standard; If all the time characteristic parameters are within the standard range, it is determined that the compensated key phase pulse signals meet the standards; If any time characteristic parameter is not within the standard range, the time drift compensation is continued until a preset maximum compensation number is reached; After it is determined that the compensated key phase pulse signals meet the standards, the following steps are performed: A1: acquiring system parameters of each shaft section, establishing translation and rotation motion equations of each shaft section based on Newton's second law and Euler's equation; the system parameters include geometric dimensions, material properties, bearing stiffness parameters, and bearing damping parameters of each shaft section; A2: Simultaneously solve the translation and rotation motion equations of all shaft segments to form a global dynamic equation set of the multi-shaft system; the global dynamic equation set is in the form of: wherein, M represents a mass matrix, C represents a damping matrix, K represents a stiffness matrix, F represents an external excitation force, generated by a key phase pulse signal and a vibration signal, and X represents a displacement vector, X represents a first order derivative of the displacement vector X with respect to time, i.e., a velocity vector, X represents a second order derivative of the displacement vector X with respect to time, i.e., an acceleration vector; A3: discretizing the multi-axis system by using the finite element analysis method, dividing the multi-axis system into N finite element units, determining the node displacement and stress-strain relationship of each finite element unit, and obtaining a finite element analysis model of the multi-axis system; After it is determined that the compensated key phase pulse signals meet the standards, the following steps are also performed: A4: convert the compensated key phase pulse signal into a rotational speed-time curve, input the rotational excitation source of the multi-shaft system into the finite element analysis model of the multi-shaft system, decompose the vibration signal into frequency domain components through Fourier transform, and map the frequency domain components to corresponding nodes of the finite element analysis model of the multi-shaft system as external excitation forces to obtain a finite element model loaded with excitation; A5: based on the finite element model loaded with excitation, solve the global dynamic equation set by using the Runge-Kutta method to obtain the displacement, velocity and acceleration of each node, and output the dynamic response of the multi-shaft system; the dynamic response of the multi-shaft system includes shaft center trajectory, bending moment distribution and bearing reaction force; the shaft center trajectory is determined based on the dynamic displacement curve of each shaft section; the bending moment distribution is calculated based on the node displacement; the bearing reaction force is calculated from the boundary conditions and the dynamic response; after determining that the compensated key phase pulse signal meets the standard, the following steps are further performed: A6: using the element shape function of the finite element model loaded with excitation, the node displacement is calculated by interpolation to obtain the strain distribution inside the shaft section, and then the stress field is obtained by combining the material constitutive relationship; A7: based on the node displacement output by the finite element model loaded with excitation, the offset of the center line of each shaft section relative to the ideal axis is calculated, the peak value is taken as the radial deviation, and the axial deviation is calculated by comprehensively considering the axial displacement difference and the inclination angle of the coupling end face; A8: integrate the radial deviation and the axial deviation calculated by the finite element model loaded with excitation to output a dynamic shaft alignment deviation report; the shaft alignment deviation report includes the deviation amount, the deviation position and the change trend over time.

2. The method of key phase pulse signal synchronous acquisition and multi-axis alignment analysis according to claim 1, characterized in that, The specific steps of S2 include: S2.1: the synchronous acquisition system simultaneously acquires the key phase pulse signal and the vibration signal according to the preset sampling frequency; the sampling frequency is determined according to the highest rotational speed of the shaft and the highest frequency component of the vibration signal; S2.2: after the key phase pulse signal and the vibration signal enter the signal conditioning circuit, they are amplified by the signal amplification module; the signal conditioning circuit includes a signal amplification module, a filter module and a signal shaping module; the amplification multiple is controlled by adjusting the resistance and capacitance parameters in the amplification circuit; S2.3: the amplified key phase pulse signal and the vibration signal enter the filter module, and the filtered key phase pulse signal and the vibration signal enter the signal shaping module again, and the filtered key phase pulse signal and the vibration signal are converted into standard square wave signals through the comparator and the Schmidt trigger to obtain the conditioned key phase pulse signal and the vibration signal; S2.4: for the conditioned key phase pulse signal, the shaft speed is calculated by measuring the time interval between adjacent pulses; the shaft speed is the ratio of 60 to the time interval between adjacent pulses; S2.5: the real-time load is obtained according to the conditioned vibration signal.

3. The method of key phase pulse signal synchronous acquisition and multi-axis alignment analysis according to claim 2, characterized in that, The specific steps of S3 include: S3.1: time characteristic analysis is performed on the conditioned key phase pulse signal, the rising edge trigger time of each pulse is extracted, and a rising edge timestamp sequence is generated; S3.2: Based on the rising edge timestamp sequence, the difference between adjacent rising edge timestamps is calculated to obtain a sequence of adjacent time intervals. If any adjacent time interval exceeds the preset percentage range of the theoretical period, it is determined as an abnormal pulse and is corrected by linear interpolation; S3.3: The corrected adjacent time interval sequence is standardized to obtain a standardized time interval sequence; S3.4: The coefficient of variation, skewness and kurtosis of the standardized time interval sequence are calculated using the sliding window statistical method to quantify the dispersion and distribution of the standardized time interval sequence; S3.5: The frequency domain analysis is performed on the standardized time interval sequence to extract the amplitude ratio of the fundamental frequency and the second harmonic, identify the shaft misalignment characteristics, and obtain the harmonic ratio; S3.6: The autocorrelation function of the standardized time interval sequence is calculated to analyze its autocorrelation characteristics; S3.7: The coefficient of variation, skewness, kurtosis, harmonic ratio and autocorrelation characteristics are integrated to form the drift characteristics; S3.8: The pre-constructed autoregressive moving average model is loaded and trained with the drift characteristics as input to generate a time drift monitoring model.

4. The method of key phase pulse signal synchronous acquisition and multi-axis alignment analysis according to claim 3, characterized in that, The specific steps of S5 include: S5.1: Collecting experimental data and preprocessing the experimental data; the experimental data includes system running state and performance indicators under different shaft alignment deviation, real-time shaft speed and load conditions; S5.2: Loading the pre-constructed nonlinear regression model and training the nonlinear regression model using the preprocessed experimental data. By adjusting the parameters of the nonlinear regression model, the error between the predicted value and the actual value of the nonlinear regression model is minimized to obtain a trained nonlinear regression model; S5.3: The trained nonlinear regression model is used as a nonlinear tolerance threshold model, and the real-time shaft alignment deviation, real-time shaft speed and load are input to output the allowable alignment deviation threshold.

5. The method of key phase pulse signal synchronous acquisition and multi-axis alignment analysis according to claim 4, characterized in that, The specific steps of S6 include: S6.1: Comparing the calculated shaft alignment deviation with the allowable alignment deviation threshold output by the nonlinear tolerance threshold model; If the radial deviation and axial deviation of the shaft alignment deviation are both less than the corresponding allowable alignment deviation threshold, output the alignment qualified conclusion, record the conclusion in the system log, and end the entire analysis process; If the radial deviation or axial deviation of the shaft alignment deviation is greater than the corresponding allowable alignment deviation threshold, generate a correction suggestion according to the size and direction of the shaft alignment deviation and the structure and operation characteristics of the multi-shaft system; S6.2: Display the correction suggestion on the operation interface and trigger the re-sampling instruction to synchronize the system to re-collect the key phase pulse signal and vibration signal for further analysis and judgment.

6. The method of key phase pulse signal synchronous acquisition and multi-axis alignment analysis according to claim 5, characterized in that, The generation of the correction suggestion includes: S6.1.1: Establish a correction suggestion knowledge base according to the size and direction of the shaft alignment deviation and the structure and operation characteristics of the multi-shaft system; the correction suggestion knowledge base contains correction methods and empirical data under different shaft alignment deviation conditions; S6.1.2: When it is detected that the shaft alignment deviation exceeds the allowable alignment deviation threshold, the most suitable correction method is found by matching and querying the correction suggestion knowledge base according to the size and direction of the shaft alignment deviation. S6.1.3: When the correction method in the correction suggestion knowledge base does not match, then generate a correction suggestion according to the change trend of the axis alignment deviation and the running state of the system; S6.1.4: The generated correction suggestion is sorted and optimized, and displayed on the operation interface.

Images