A method for detecting acceleration vibration of a continuous casting crystallizer by using an optical fiber

Abstract

The application provides a kind of optical fiber acceleration vibration detection method for continuous casting crystallizer, it is related to continuous casting crystallizer vibration detection technical field, optical fiber acceleration probe is respectively arranged in three orthogonal directions of continuous casting crystallizer, method includes: by optical fiber acceleration probe acquisition acceleration signal, obtains acceleration signal set, acceleration signal set includes at least one acceleration signal corresponding to each direction of continuous casting crystallizer in three orthogonal directions;Acceleration signal set is handled to obtain the vibration information of continuous casting crystallizer in three orthogonal directions;According to vibration information, judge whether abnormal vibration condition appears;When abnormal vibration condition appears, vibration abnormal type is identified according to vibration information;According to vibration abnormal type, generate detection report.The vibration detection method provided in the application realizes three-dimensional vibration detection using optical fiber acceleration probe, can effectively adapt to the harsh environment of continuous casting site, and provides direct basis for maintenance through the classification identification of abnormal vibration.

Description

Technical Field

[0001] This invention belongs to the field of vibration detection technology for continuous casting crystallizers, and specifically relates to a fiber optic acceleration vibration detection method for continuous casting crystallizers. Background Technology

[0002] In continuous casting, the crystallizer, as the core equipment of the continuous casting machine, directly affects the surface quality and internal structure of the cast billet through its vibration parameters (including amplitude, frequency, and waveform). Driven by a vibration device, the crystallizer performs periodic movements according to a set amplitude, frequency, and waveform pattern to prevent the cast billet from sticking to the crystallizer wall and causing leakage. However, the crystallizer operates in a harsh environment of high temperature, high humidity, and strong electromagnetic interference. During operation, its vibration system may experience abnormal vibrations due to mechanical component wear, hydraulic system leaks, transmission mechanism malfunctions, or misalignment. These abnormal vibrations manifest as abnormal vibration frequency, amplitude, or direction. If these abnormal vibrations are not detected and accurately identified in a timely manner, they will lead to deeper vibration marks on the cast billet, deterioration of surface quality, and even production accidents such as sticking and leakage.

[0003] Currently, crystallizer vibration detection mainly employs piezoelectric accelerometers or laser displacement sensors. Piezoelectric accelerometers are susceptible to signal interference and lack long-term stability in high-temperature and strong electromagnetic interference environments. While laser displacement sensors offer non-contact measurement, they are significantly affected by factors such as on-site dust and space limitations of the vibration table. Most existing detection methods can only obtain basic parameters such as vibration amplitude and frequency, lacking the ability to systematically identify abnormal vibration types, making it difficult to directly provide maintenance personnel with targeted fault diagnosis information. Summary of the Invention

[0004] In view of the above-mentioned defects or deficiencies in the prior art, this application provides a method for detecting fiber optic acceleration vibration in a continuous casting mold, wherein fiber optic acceleration probes are respectively installed in three orthogonal directions of the continuous casting mold, and the method includes: Acceleration signals are acquired by the fiber optic accelerometer probe to obtain an acceleration signal set, which includes at least one acceleration signal corresponding to each of the three orthogonal directions of the continuous casting crystallizer. The acceleration signal set is processed to obtain vibration information of the continuous casting crystallizer in three orthogonal directions, the vibration information including amplitude and frequency; Determine whether abnormal vibration has occurred based on the vibration information; When abnormal vibration occurs, the type of vibration abnormality is identified based on the vibration information. A detection report is generated based on the type of vibration anomaly.

[0005] According to the technical solution provided by the present invention, the step of determining whether abnormal vibration occurs based on vibration information includes: The instantaneous amplitude sequence of vibration signals in three orthogonal directions is extracted, and the instantaneous amplitudes at each time point are connected in order of sampling time to construct a three-dimensional vibration phase trajectory curve. The curvature and deflection of the phase trajectory at each sampling point are calculated. Calculate the curvature deviation between the curvature and the theoretical curvature at each sampling point, and the deflection deviation between the deflection and the theoretical deflection. Also calculate the cumulative value of the curvature deviation and the cumulative value of the deflection within one vibration cycle. The theoretical curvature and theoretical deflection are determined by the theoretical vibration trajectory of the crystallizer under the current operating parameters. Abnormal vibration is defined as the cumulative value of curvature deviation exceeding the first threshold or the cumulative value of deflection deviation exceeding the second threshold. When both exceed the corresponding threshold, it is defined as severe abnormal vibration.

[0006] According to the technical solution provided by the present invention, the step of identifying the type of vibration anomaly based on vibration information includes: The time-domain and frequency-domain features of vibration signals in three orthogonal directions during the abnormal vibration period are extracted, and the distortion direction feature vector of the abnormal vibration in three-dimensional space is constructed based on the curvature deviation and deflection deviation. The time-domain features include at least peak value, root mean square value, kurtosis and waveform factor, and the frequency-domain features include at least the main frequency amplitude, harmonic energy ratio and sideband amplitude. The extracted time-domain features, frequency-domain features, and distortion direction feature vectors are input into a pre-established anomaly identification model, which outputs vibration anomaly types, including mechanical component wear anomalies, hydraulic system leakage anomalies, transmission mechanism failure anomalies, and misalignment anomalies.

[0007] According to the technical solution provided by the present invention, after determining whether an abnormal vibration has occurred based on the vibration information, the method further includes: Calculate the distribution ratio of vibration energy in each direction at the fundamental frequency and each harmonic frequency to construct a three-dimensional energy distribution matrix; Calculate the norm deviation between the current energy distribution matrix and the reference energy distribution matrix under normal operating conditions; When the norm deviation exceeds a preset threshold, the spectral structure is determined to be abnormal.

[0008] According to the technical solution provided by the present invention, after determining whether an abnormal vibration has occurred based on the vibration information, the method further includes: Calculate the instantaneous phase difference sequence between each pair of vibration signals in the three orthogonal directions, and calculate the mean and standard deviation of each phase difference sequence within one vibration cycle; When any mean deviates from the theoretical phase difference by more than the first deviation threshold, it is determined that there is a phase synchronization abnormality between the corresponding direction pairs. When any standard deviation exceeds the second deviation threshold, it is determined that there is a phase jitter abnormality between the corresponding direction pairs.

[0009] According to the technical solution provided by the present invention, at least two fiber optic accelerometer probes are provided in each of the three orthogonal directions, and the at least two fiber optic accelerometer probes in the same direction are installed at a predetermined distance apart along that direction, and the acceleration signal set includes at least two acceleration signals corresponding to each direction; Before processing the acceleration signal set, the method further includes: Verify the validity of at least two acceleration signals in the same direction and discard invalid signals; When there are two valid signals in the same direction, the composite acceleration signal in that direction is calculated based on the two valid signals; when there are one or zero valid signals in the same direction, fault tolerance processing is performed. The acceleration signal set is updated using the synthesized acceleration signals from each direction or the results of fault-tolerant processing.

[0010] According to the technical solution provided by the present invention, the step of calculating the composite acceleration signal in that direction based on two valid signals includes: The installation distance between two effective probes along this direction is obtained, the cross-correlation function between the two effective signals is calculated to determine the time delay between them, and the propagation speed of the vibration wave along this direction is calculated based on the installation distance and the time delay. The two valid signals are weighted and synthesized using the propagation speed as the weight, with the signal having a higher propagation speed having a higher weight. The spatial gradient of vibration acceleration in this direction is calculated based on the amplitude difference between the two valid signals and the installation spacing. When the spatial gradient exceeds a preset threshold, it is determined that there is an abnormality in the vibration field in this direction and an early warning is generated.

[0011] According to the technical solution provided by the present invention, the fault-tolerant processing includes: When the number of valid signals in the same direction is one, and the probe corresponding to the valid signal is the only valid probe, the valid signal is used as the final signal in that direction and directly participates in subsequent processing, while generating a single probe working prompt message. When the number of valid signals in the same direction is zero, extrapolation prediction is performed using the synthetic acceleration signal in that direction at the previous moment to obtain the replacement signal at the current moment, and an alarm message indicating that all probes have failed is generated.

[0012] According to the technical solution provided by the present invention, at least two fiber optic acceleration probes are provided in each of the three orthogonal directions, including a sensing fiber optic probe and a reference fiber optic probe. The first response signal output by the sensing fiber optic probe simultaneously includes wavelength drift components caused by vibration and temperature. The reference fiber optic probe and the sensing fiber optic probe use the same fiber grating parameters and are installed in the same position. The second response signal output by the reference fiber optic probe only includes wavelength drift components caused by temperature. Before processing the acceleration signal set, the method further includes: Simultaneously acquire the first response signal and the second response signal, and establish a temperature offset reference value under the current temperature conditions based on the second response signal; The first response signal is differentially compensated in real time using the temperature offset reference value to obtain a vibration acceleration signal that eliminates the influence of temperature. The acceleration signal set is updated using the compensated and corrected vibration acceleration signal.

[0013] According to the technical solution provided by the present invention, the real-time differential compensation of the first response signal using a temperature offset reference value includes: Acquire the first wavelength drift of the sensing fiber optic probe between the current time and the initial time, and the second wavelength drift of the reference fiber optic probe between the current time and the initial time; Using the second wavelength drift as a reference for temperature-induced wavelength drift, the second wavelength drift is subtracted from the first wavelength drift to obtain the wavelength drift caused by pure vibration.

[0014] Compared with existing technologies, the advantages of this invention are as follows: By setting fiber optic acceleration probes in three orthogonal directions of the continuous casting crystallizer, and using fiber optic sensing to replace traditional piezoelectric or laser sensors, the defects of signal interference in high-temperature and strong electromagnetic interference environments are effectively overcome, and the influence of on-site dust and space limitations on measurement is avoided, thus improving the environmental adaptability and long-term stability of the detection system; by collecting acceleration signals in three orthogonal directions and extracting the amplitude and frequency of each direction, complete vibration information of the crystallizer in three-dimensional space can be obtained simultaneously, overcoming the limitation of existing methods that can only detect vibration in a single direction; by comprehensively judging abnormal vibration based on three-dimensional vibration information and identifying the type of vibration abnormality when an abnormality occurs, the system is upgraded from single-point alarm to systematic abnormality classification and diagnosis, providing maintenance personnel with direct fault judgment basis, and by generating a detection report containing the abnormality type, the intelligent level of vibration detection and fault response efficiency of the continuous casting crystallizer are effectively improved. Attached Figure Description

[0015] Other features, objects, and advantages of this application will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings: Figure 1 This is a flowchart of the steps of the fiber optic acceleration vibration detection method for a continuous casting crystallizer provided in Example 1; Figure 2 This is a flowchart illustrating the steps in Example 2 that differ from those in Example 1; Figure 3 This is a flowchart illustrating the steps in Example 3 that differ from those in Example 1; Figure 4 This is a flowchart of the steps in Example 4 that differ from those in Example 1; Figure 5 This is a flowchart of the steps in Example 5 that differ from those in Example 1. Detailed Implementation

[0016] The present application will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and not intended to limit it. Furthermore, it should be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings.

[0017] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0018] Example 1 As mentioned in the background section, please refer to the relevant technical issues. Figure 1 This embodiment proposes a fiber optic acceleration vibration detection method for continuous casting crystallizers. Fiber optic acceleration probes are installed in three orthogonal directions of the continuous casting crystallizer. The method includes: S100: Acceleration signals are acquired through the fiber optic accelerometer probe to obtain an acceleration signal set, which includes at least one acceleration signal corresponding to each of the three orthogonal directions of the continuous casting crystallizer.

[0019] Specifically, in step S100, three fiber Bragg grating (FBG) accelerometers are fixedly installed on the outer wall of the continuous casting crystallizer or on the vibration table, corresponding to the three orthogonal directions of the X, Y, and Z axes, respectively. The three sensors are connected to the signal processing unit via optical fibers and share the same light source and demodulator. The FBG accelerometers operate based on the inertial mass-FBG coupling principle: the mass block inside the sensor generates inertial force under the action of vibration acceleration, causing axial strain in the fiber grating, resulting in a shift in the grating's center wavelength. The demodulator measures the wavelength shift of each channel in real time, synchronously acquiring acceleration signals in the three directions according to a preset sampling frequency (e.g., 2000Hz), and after analog-to-digital conversion, forms digitized acceleration time-domain signals, denoted as follows: , , The acceleration signal in each direction is a time series, containing data from at least one complete vibration cycle, thus forming an acceleration signal set. .

[0020] S200: Process the acceleration signal set to obtain vibration information of the continuous casting crystallizer in three orthogonal directions, the vibration information including amplitude and frequency.

[0021] Specifically, in step S200, the acquired acceleration signals in the three directions are preprocessed, including: using a bandpass filter to filter out high-frequency noise and power frequency interference, and simultaneously performing baseline correction to eliminate zero drift. The preprocessed acceleration signals in each direction are then subjected to time-frequency domain transformation to extract the vibration frequency components. Specifically, a Fast Fourier Transform (FFT) can be used to convert the time-domain acceleration signal into a spectrum, and the frequency corresponding to the peak value of the dominant frequency in the spectrum is identified as the vibration frequency in that direction. Simultaneously, based on the integral relationship between the acceleration signal and the displacement signal, the acceleration signal is integrally integrated twice to obtain the vibration displacement signal, and the peak value of the displacement signal is then extracted as the vibration amplitude in that direction. Taking the X-axis direction as an example, for... The displacement signal is obtained by performing a second integration. The peak value within one period is taken as the amplitude in the X-axis direction. ;right Perform an FFT transform and take the frequency corresponding to the maximum amplitude of the spectrum as the frequency in the X-axis direction. Similarly, the acceleration signals in the Y and Z axes are processed to obtain the amplitude in each direction. , and frequency , This constitutes complete vibration information in three orthogonal directions.

[0022] S300: Determine whether abnormal vibration has occurred based on the vibration information.

[0023] Specifically, in step S300, based on the vibration information (including amplitude and frequency in each direction) extracted in step S200 in the three orthogonal directions, a comprehensive judgment is made as to whether there is abnormal vibration in the continuous casting mold. The judgment is based on, but is not limited to, whether the amplitude in each direction exceeds the normal operating range, whether the frequency in each direction deviates from the theoretical set value, and whether the phase relationship and trajectory shape between the vibrations in the three directions are distorted.

[0024] S400: When abnormal vibration occurs, the type of vibration abnormality is identified based on the vibration information.

[0025] Specifically, in step S400, when step S300 determines that abnormal vibration has occurred in the continuous casting mold, the type of abnormal vibration is further identified. The identification is based on the time-domain waveform characteristics, spectral distribution characteristics, and vibration trajectory distortion characteristics in three-dimensional space of the vibration signals in each direction during the period of abnormality. By extracting the above multi-dimensional features and matching them with pre-established fault modes, the abnormal vibration can be classified into specific fault types, such as abnormal wear of mechanical parts, abnormal leakage of hydraulic system, abnormal failure of transmission mechanism, or abnormal misalignment.

[0026] S500: Generate a detection report based on the described vibration anomaly type.

[0027] Specifically, in step S500, a formatted inspection report is automatically generated based on the vibration anomaly type identified in step S400. The inspection report includes at least: a timestamp of the abnormal vibration occurrence, the duration of the anomaly, amplitude and frequency data in three orthogonal directions during the anomaly, the severity level of the abnormal vibration (ordinary or severe), the specific anomaly type identified, and maintenance recommendations for that anomaly type. For example, if the identified anomaly type is "hydraulic system leakage anomaly," the inspection report may provide a maintenance suggestion to "check hydraulic lines and sealing elements." The inspection report can be output in text file, table, or graphical interface format for reference by on-site operators or equipment maintenance personnel to take timely maintenance measures.

[0028] Further, in step S300, the step of determining whether abnormal vibration has occurred based on the vibration information includes the following steps S310-S360: S310: Extract the instantaneous amplitude sequence of vibration signals in three orthogonal directions.

[0029] Specifically, in step S310, instantaneous amplitude sequences are extracted from the vibration signals in each direction processed in step S200. For acceleration signals... , , The instantaneous amplitude envelope at each moment is calculated using the Hilbert Transform. Taking the X-axis direction as an example, for... The analytic signal is obtained by performing a Hilbert transform, and the magnitude of the analytic signal is the instantaneous amplitude at that moment. By processing the acceleration signals in the Y and Z axes in this manner, instantaneous amplitude sequences are obtained respectively. and The three instantaneous amplitude sequences are synchronized in time, with each sampling time... Each corresponds to a set of three-dimensional instantaneous amplitudes ( , , The instantaneous amplitude sequence reflects the vibration intensity of the crystallizer in three directions at each instant, providing basic data for subsequent construction of the three-dimensional spatial trajectory.

[0030] S320: Based on the instantaneous amplitudes in three orthogonal directions at the same moment, instantaneous vibration state points in three-dimensional space are constructed, and the instantaneous vibration state points at each moment are connected sequentially in order of sampling time to form a three-dimensional vibration phase trajectory curve.

[0031] Specifically, in step S320, the three orthogonal directions are regarded as the three coordinate axes of three-dimensional space, with the instantaneous amplitude in the X-axis direction as the reference value. The x-coordinate of a spatial point, expressed as the instantaneous amplitude along the Y-axis. As the y-coordinate, the instantaneous amplitude in the Z-axis direction As the z-coordinate. For each sampling time... ,Will( , , A point is defined as an instantaneous vibration state point in three-dimensional space. The spatial position of this point represents the combined state of the crystallizer's vibration amplitude in three directions at that moment. Connecting the instantaneous vibration state points at each sampling time sequentially forms a curve in three-dimensional space, called the three-dimensional vibration phase trajectory curve. This curve comprehensively describes the changes in the three-dimensional vibration state of the crystallizer within one vibration cycle: each point on the curve represents the vibration state at a given moment, and the direction and shape of the curve reflect the phase relationship and amplitude ratio between the vibrations in the three directions. Compared to observing vibration signals in only one direction, the three-dimensional phase trajectory can intuitively present the overall spatial behavior of the crystallizer vibration; any change in the phase relationship or amplitude ratio between directions will be reflected in the trajectory shape. For example, when the vibrations in the three directions are completely synchronized, the phase trajectory appears as a thin, elongated curve near a spatial diagonal; when there is a phase difference between the vibrations in each direction, the phase trajectory exhibits a spatial ellipse or a more complex spatial closed curve shape.

[0032] S330: Calculate the curvature and torsion of the three-dimensional vibration phase trajectory curve at each sampling point. The calculation formula is as follows:

[0033]

[0034] in For curvature, For torsion, The position vector of the three-dimensional vibration phase trajectory curve. , , They are respectively The first, second, and third derivatives.

[0035] Specifically, in step S330, curvature This describes the curvature of the three-dimensional vibration phase trajectory in space. For a spatial curve, a greater curvature indicates a more pronounced curvature at that point; zero curvature indicates that the curve extends in a straight line at that point. Torsion This describes the degree to which the three-dimensional vibration phase trajectory deviates from a planar curve, i.e., the degree of torsion of the curve. A larger torsion indicates that the curve is more "twisted" at that point, deviating more from planar motion; zero torsion indicates that the curve lies within a plane at that point. In the application scenario of crystallizer vibration detection, the physical meaning of curvature and torsion is very clear: during normal operation of a continuous casting crystallizer, its vibration in three directions is controlled by the same hydraulic drive system, with a definite amplitude ratio and phase relationship. Therefore, its three-dimensional phase trajectory is a spatial closed curve with stable curvature and torsion characteristics. When wear of the guide mechanism causes a decrease in the constraint stiffness in a certain direction, the curvature of the trajectory in that direction will change; when abnormal response of the hydraulic system causes phase lag in a certain direction, the torsion of the trajectory will change, and the torsion will change. By calculating the curvature and deflection at each sampling point, the analysis of the vibration signal's "amplitude" and "frequency" is upgraded to the analysis of the vibration trajectory's "morphology," that is, from "how big and how fast the vibration is" to "how much the vibration trajectory is bent and twisted," thus enabling more sensitive capture of trajectory morphology distortions that are difficult to reflect by traditional features.

[0036] S340: Calculate the curvature deviation between the curvature at each sampling point and the theoretical curvature of the theoretical vibration trajectory of the continuous casting crystallizer under the current operating parameters, as well as the deflection deviation between the deflection and the theoretical deflection.

[0037] Specifically, in step S340, the theoretical vibration trajectory is first calculated based on the current operating parameters of the continuous casting mold (including casting speed, vibration frequency, and the set vibration waveform type, such as sine wave or non-sine wave). The theoretical vibration trajectory reflects the three-dimensional phase trajectory shape that the mold should exhibit under ideal conditions (no faults, all components in good working order), and its theoretical curvature... and theoretical torsion The actual curvature calculated in step S330 can be obtained in advance through the kinematic model of the crystallizer. With theoretical curvature The curvature deviation is obtained by subtracting the values ​​point by point at the same sampling point and taking the absolute value. Similarly, the actual torsion... With theoretical torsion Subtracting each point and taking the absolute value yields the torsion deviation. Curvature deviation and torsion deviation It reflects the degree of deviation of the actual vibration trajectory from the theoretical trajectory at each instant. The larger the deviation value, the more severe the trajectory distortion at that moment.

[0038] S350: Calculate the cumulative values ​​of curvature deviation and deflection deviation within one vibration cycle.

[0039] Specifically, in step S350, the period corresponding to the current vibration frequency of the crystallizer is used. As the integration window, the curvature deviation calculated in step S340 is used. and torsion deviation Integrating over a complete vibration cycle (for discrete sampled data, numerical integration is used to accumulate the values) yields the cumulative periodic value of the curvature deviation. The cumulative value of the periodic deviation of the torsion Using cumulative values ​​over a period of time, rather than instantaneous values ​​at a single point, as the basis for judgment can effectively avoid misjudgments caused by instantaneous deviation spikes due to accidental noise interference, making the judgment results more stable and reliable. The cumulative amount of deviation within a vibration cycle represents the total degree of distortion of the overall trajectory shape within that cycle, and is a comprehensive indicator for measuring whether the crystallizer's operating state deviates from the normal level.

[0040] S360: When the cumulative value of curvature deviation exceeds the first preset threshold or the cumulative value of deflection exceeds the second preset threshold, it is determined that an abnormal vibration has occurred; when both the cumulative value of curvature deviation and the cumulative value of deflection exceed their respective thresholds, it is determined that a serious abnormal vibration has occurred.

[0041] Specifically, in step S360, the cumulative curvature deviation value calculated in step S350 is compared with a preset first threshold, and the cumulative deflection deviation value is compared with a preset second threshold. The first and second thresholds can be determined based on the statistical distribution of the cumulative curvature and deflection deviation values ​​measured multiple times during normal operation of the continuous casting mold. For example, three times the standard deviation of the average value under normal operating conditions can be used as the threshold, or it can be set based on engineering experience. When the cumulative curvature deviation exceeds the first threshold or the cumulative deflection deviation exceeds the second threshold, it indicates that the morphological distortion of the actual vibration trajectory has exceeded the normal allowable range, and is judged as an abnormal vibration situation. When both the cumulative curvature deviation exceeding the first threshold and the cumulative deflection deviation exceeding the second threshold are simultaneously met, it indicates that the vibration trajectory has undergone significant distortion in both bending and torsional dimensions, and the mold vibration system may have a relatively serious complex fault, and is judged as a serious abnormal vibration situation. This grading judgment mechanism provides on-site operators with an intuitive judgment of the severity of the abnormality—ordinary abnormalities indicate the need for attention, while serious abnormalities indicate the need for immediate shutdown and maintenance, thus effectively guiding maintenance decisions.

[0042] Further, in step S400, the identification of vibration anomaly types based on vibration information includes the following steps S410-S430: S410: Extract the time-domain and frequency-domain characteristic parameters of the vibration signal in three orthogonal directions during the period of abnormal vibration occurrence. The time-domain characteristic parameters include at least the vibration peak value, root mean square value, kurtosis and waveform factor. The frequency-domain characteristic parameters include at least the main frequency amplitude, harmonic energy ratio and sideband amplitude.

[0043] S420: Construct the distortion direction feature vector of abnormal vibration in three-dimensional space based on curvature deviation and deflection deviation: .

[0044] Specifically, in step S420, the curvature deviation and torsion deviation While these values ​​can reflect the degree to which the trajectory shape deviates from the theoretical value, they are all scalar quantities and cannot describe the directional characteristics of the deviation in three-dimensional space. To compensate for this deficiency, this step constructs a distortion direction feature vector. .in, and These are the unit vectors representing the curvature deviation and torsion deviation in three-dimensional space, respectively. Their directions are determined by the geometric relationship between the tangent vector, principal normal vector, and binormal vector of the three-dimensional vibration phase trajectory at that point. Distortion direction eigenvector The three components represent: The cumulative intensity of curvature deviation in three-dimensional space is the magnitude of the curvature deviation values ​​at each moment, which is the result of vector accumulation along the unit vector of its direction within the integration window. The cumulative intensity of the torsion deviation in three-dimensional space; and This represents the cumulative intensity of the difference between curvature deviation and deflection deviation. This feature vector expands the magnitude information of the deviation to include both magnitude and direction information. Different fault types lead to different spatial direction characteristics of trajectory distortion. For example, mechanical wear often leads to enhanced bending distortion in a specific direction, while hydraulic leakage often leads to an increase in torsional distortion components, thus providing a more refined basis for anomaly type identification.

[0045] S430: Input the extracted time-domain feature parameters, frequency-domain feature parameters, and the distortion direction feature vector into the pre-established anomaly identification model, and output the vibration anomaly type, which includes mechanical component wear anomaly, hydraulic system leakage anomaly, transmission mechanism failure anomaly, and misalignment anomaly.

[0046] Specifically, in step S430, the time-domain and frequency-domain feature parameters extracted in step S410 are combined with the distortion direction feature vector constructed in step S420 to form a multi-dimensional feature vector, which is then input into the pre-trained anomaly recognition model. The anomaly recognition model can employ machine learning models such as support vector machines, random forests, or lightweight neural networks. Its training process is as follows: Under normal crystallizer operation and various known fault conditions, a large amount of data is collected using this method, and the aforementioned multi-dimensional features are extracted to construct a labeled training sample set. The model is trained using the multi-dimensional features as input and the operating state type as the label. After model training is complete, the real-time extracted feature vector is input into the model, and the model outputs the corresponding anomaly type, including mechanical component wear anomalies, hydraulic system leakage anomalies, transmission mechanism failure anomalies, and misalignment anomalies. Through multi-dimensional feature fusion and machine learning classification, automatic and accurate identification of anomaly types is achieved.

[0047] Example 2 Based on the above embodiment 1, please refer to Figure 2 This embodiment provides another method for detecting fiber optic acceleration vibration in continuous casting crystallizers. The same content as in Embodiment 1 will not be repeated here; the difference lies in: In this embodiment, after step S300 and before step S400, the following steps S371-S374 are also included: S371: Based on the vibration information in three orthogonal directions, calculate the energy distribution ratio of vibration energy in each direction at the fundamental frequency and its harmonics of the continuous casting crystallizer.

[0048] Specifically, in step S371, based on the vibration signals in the three orthogonal directions obtained in step S200, a Fast Fourier Transform is performed on the acceleration time-domain signal in each direction to obtain the spectrum for each direction. The fundamental frequency component corresponding to the current operating frequency of the crystallizer is then identified from the spectrum. and extract the fundamental frequency. and its second harmonic Third harmonic Until Frequency multiplier Spectral amplitude at that location The value of is determined based on actual needs, and typically 5 to 10 is sufficient to cover the main harmonic energies. For each direction, the square of the amplitude at each harmonic is calculated as the energy value at that frequency. Then, the sum of the energy values ​​at all harmonics in that direction is calculated. The energy distribution ratio at each harmonic is obtained by dividing the energy value at each harmonic by the total energy value. Taking the X-axis direction as an example, the energy proportion at the fundamental frequency is... The energy percentage at the second harmonic is And so on. Similarly, processing the Y-axis and Z-axis directions yields the energy distribution ratios at each octave. These energy distribution ratios reflect the concentration and dispersion of vibrational energy along the frequency dimension; different fault types will cause energy to redistribute to different degrees at different octaves.

[0049] S372: Constructing a three-dimensional energy distribution matrix:

[0050] in, express Direction in The percentage of energy at the first harmonic.

[0051] Specifically, in step S372, the energy distribution ratios of the three orthogonal directions at each harmonic frequency calculated in step S371 are organized in matrix form. The rows of the matrix correspond to the three orthogonal directions, namely the X-axis, Y-axis, and Z-axis directions; the columns of the matrix correspond to the fundamental frequency and each harmonic frequency, that is, the first column corresponds to the fundamental frequency. The second column corresponds to the second harmonic. The third column corresponds to the third harmonic. And so on up to the 1st Column correspondence Frequency multiplier Each element in the matrix express Direction in The energy percentage at the first harmonic, of which Choose X, Y, or Z. Take 1 to This three-dimensional energy distribution matrix comprehensively describes the distribution of crystallizer vibration energy in both spatial and frequency dimensions, serving as a digital representation of the spectral structure of the crystallizer vibration system. Organizing the energy distribution data in matrix form facilitates subsequent matrix operations and quantitative comparisons.

[0052] S373: Calculate the norm deviation between the current energy distribution matrix and the reference energy distribution matrix under normal operating conditions of the continuous casting mold:

[0053] in, This is the baseline energy distribution matrix under normal operating conditions of the crystallizer.

[0054] Specifically, in step S373, the reference energy distribution matrix The standard energy distribution matrix, obtained by averaging multiple measurements after the continuous casting crystallizer has been successfully installed and commissioned, and is in good operating condition, following the same procedure as steps S371 and S372, represents the spectral structure fingerprint of the crystallizer in a healthy state. The currently constructed energy distribution matrix... Compared with the baseline energy distribution matrix By subtracting element by element, we obtain the difference matrix. The norm of the difference matrix is ​​calculated by taking the square root of the sum of the squares of all elements in the difference matrix, yielding the norm deviation value. This norm deviation value is a scalar representing the overall degree of difference between the current spectral structure and the baseline spectral structure under healthy conditions. The larger the norm deviation value, the more significant the change in the frequency distribution of the current vibrational energy compared to normal operation, and the more severe the alteration in the spectral structure of the crystallizer vibration system.

[0055] S374: When the norm deviation exceeds the preset energy distortion threshold, it is determined that the spectral structure of the continuous casting crystallizer vibration system has changed abnormally. This determination result is used to verify the judgment result of step S300.

[0056] Specifically, in step S374, the norm deviation value calculated in step S373 is compared with a preset energy distortion threshold. The energy distortion threshold can be determined based on the statistical distribution of norm deviation values ​​measured multiple times under normal operating conditions of the crystallizer. For example, three times the standard deviation of the average norm deviation under normal operating conditions can be used as the threshold. When the norm deviation value exceeds the energy distortion threshold, it indicates that the spectral distribution of the current vibration energy has significantly deviated from a healthy state, and it is determined that the spectral structure of the crystallizer vibration system has undergone abnormal changes. This determination result is independent of the judgment result based on the three-dimensional phase trajectory curvature and torsion analysis in step S300. The two judgment bases are different: step S300 focuses on the spatial geometric distortion of the vibration trajectory, while this step focuses on the changes in the distribution structure of vibration energy in the frequency domain. When the two judgment results are consistent, the reliability of the anomaly judgment is mutually verified; when they are inconsistent, it indicates that there may be a fault mode that is difficult to capture in a single dimension, requiring further comprehensive analysis. Through this multi-dimensional mutual verification judgment mechanism, the risk of misjudgment or omission that may be caused by a single judgment basis is effectively reduced, and the accuracy and robustness of anomaly detection are improved. In other embodiments, steps S371-S374 can also be used alone as a means of judging abnormal vibration conditions.

[0057] Furthermore, after step S374, the following step S375 is also included: S375: Based on the distribution differences of the norm deviation in each direction and at each octave, locate the main contributing direction and octave order of the spectral structure anomaly, and determine the type of energy source of the vibration anomaly accordingly.

[0058] Specifically, in step S375, after step S374 determines that an abnormality has occurred in the spectral structure, a contribution analysis is further performed on the difference matrix to locate the main source causing the spectral structure abnormality. Each element in the difference matrix represents the change in the energy proportion of a certain direction at a certain harmonic frequency relative to the reference value. To quantify the contribution of each direction and each harmonic frequency to the overall norm deviation, the sum of squares of the elements in each row of the difference matrix is ​​calculated as the contribution value of that direction to the overall deviation. The direction with the largest contribution value is the main contributing direction of the spectral structure abnormality, which usually corresponds to the location of the component in the vibration system that first deteriorates or fails. Similarly, the sum of squares of the elements in each column of the difference matrix is ​​calculated as the contribution value of that harmonic frequency order to the overall deviation. The harmonic frequency order with the largest contribution value is the main contributing harmonic frequency of the spectral structure abnormality.

[0059] By analyzing the combination of the main contribution direction and the main contributing harmonic order, the type of energy source of vibration anomalies can be preliminarily determined. When the norm deviation is mainly contributed by energy changes at the fundamental frequency, it indicates that the external excitation force on the crystallizer has changed, possibly pointing to global factors such as changes in driving torque or pulling speed. When the norm deviation is mainly contributed by energy changes at the second harmonic, it indicates that the vibration system exhibits nonlinear response characteristics, possibly pointing to local faults such as increased clearance in the guide mechanism or loosening of connections. When the norm deviation is mainly contributed by energy changes at the third harmonic and higher harmonics, it indicates that the system has strong impact excitation, possibly pointing to serious faults such as bearing damage or abnormal gear meshing. Combining the main contribution direction can further narrow down the fault location range. This analysis provides supplementary judgment information from an energy perspective for subsequent anomaly type identification.

[0060] Example 3 Based on the above embodiment 1, please refer to Figure 3 This embodiment provides another method for detecting fiber optic acceleration vibration in continuous casting crystallizers. The same content as in Embodiment 1 will not be repeated here; the difference lies in: In this embodiment, after step S300 and before step S400, the following steps S381-S384 are also included:

[0061] S381: Calculate the instantaneous phase difference sequences between each pair of vibration signals in the three orthogonal directions to obtain three phase difference sequences. , and .

[0062] Specifically, in step S381, based on the vibration signals in the three orthogonal directions obtained in step S200, a Hilbert transform is performed on the signals in each direction to extract the instantaneous phase at each moment. For the X-axis direction signal, its instantaneous phase is denoted as... For signals in the Y-axis direction, their instantaneous phase is denoted as... For the Z-axis direction signal, its instantaneous phase is denoted as... Calculate the instantaneous phase difference between each pair of the three directions: the phase difference sequence between the X-axis and Y-axis is as follows. Similarly, the phase difference sequence between the Y-axis and Z-axis can be obtained. and the phase difference sequence between the Z-axis and the X-axis The three phase difference sequences reflect the relative phase relationships between vibrations in three directions. For crystallizer vibrations controlled by the same drive system, the vibrations in the three directions are not independent; their phase differences are determined by the geometric relationship and kinematic characteristics of the transmission chain, and have definite theoretical values. When any link in the transmission chain malfunctions, these phase differences will change accordingly. Therefore, the phase difference sequence is an important basis for diagnosing transmission chain faults.

[0063] S382: Calculate the mean and standard deviation of each phase difference sequence over one vibration period.

[0064] Specifically, in step S382, using the period corresponding to the current vibration frequency of the crystallizer as a window, the mean and standard deviation of the three phase difference sequences within one vibration cycle are calculated respectively. The mean phase difference reflects the average phase difference between two directions within one cycle; the standard deviation reflects the degree of fluctuation of the phase difference around the mean within one cycle. Similarly, the mean and standard deviation of the remaining two sets of phase difference sequences are calculated. The mean and standard deviation characterize the phase difference from two different perspectives: static and dynamic, respectively. The mean represents the average level of the phase difference, and its deviation from the theoretical value reflects a systematic phase shift in the transmission chain; the standard deviation represents the stability of the phase difference, and its increase reflects phase jitter caused by gaps or looseness in the transmission chain.

[0065] S383: When the mean of any phase difference sequence deviates from its theoretical phase difference value by more than the first phase deviation threshold, it is determined that there is a phase synchronization anomaly between the corresponding direction pairs.

[0066] Specifically, in step S383, the mean value of each phase difference sequence calculated in step S382 is compared with the corresponding theoretical phase difference value. The theoretical phase difference value is determined by the geometric relationship and kinematic characteristics of the crystallizer transmission chain and can be pre-calculated using the kinematic model of the crystallizer. Taking the phase difference between the X-axis and Y-axis as an example, when the mean value of this phase difference deviates from its theoretical value by more than the first phase deviation threshold, it indicates that the average phase relationship between the X-axis and Y-axis has deviated from the design value, and it is determined that there is a phase synchronization anomaly between the X-axis and Y-axis. The first phase deviation threshold can be determined based on the fluctuation range of the mean value of the phase difference measured multiple times under normal operating conditions of the crystallizer. The phase synchronization anomaly reflects a systematic shift in the phase of a certain link in the transmission chain, such as the phase deviation of the eccentric shaft or the positioning error of the cam mechanism. Based on which specific direction pair has the phase synchronization anomaly, the link where the phase shift occurs can be preliminarily determined. For example, if only the phase difference between the X-axis and Y-axis is abnormal while other directions are normal, it may point to the transmission connection link between the X-axis and Y-axis.

[0067] S384: When the standard deviation of any phase difference sequence exceeds the second phase deviation threshold, it is determined that there is a phase jitter anomaly between the corresponding direction pairs.

[0068] Specifically, in step S384, the standard deviation of each phase difference sequence calculated in step S382 is compared with a preset second phase deviation threshold. Taking the phase difference between the X-axis and Y-axis as an example, when the standard deviation of this phase difference exceeds the second phase deviation threshold, it indicates that the phase difference between the X-axis and Y-axis fluctuates violently within one cycle, and it is determined that there is a phase jitter abnormality between the X-axis and Y-axis. The second phase deviation threshold can be determined based on the statistical distribution of the standard deviation of the phase difference measured multiple times under normal operating conditions of the crystallizer. The phase jitter abnormality reflects factors such as gaps or slack constraints in the transmission chain that cause instability in the phase relationship. For example, excessive gaps at the connecting rod hinge will increase the fluctuation of the phase difference in each cycle, and slack constraints on the guide plate will cause random jitter in the phase difference. The determination of phase synchronization abnormality and phase jitter abnormality focuses on the static accuracy and dynamic stability of the phase relationship, respectively. The two complement each other and can more comprehensively reflect the health status of the transmission chain.

[0069] Furthermore, step S384 is followed by step S385: S385: Based on the direction pair combination corresponding to the phase synchronization abnormality and the phase jitter abnormality, determine the fault location of the transmission chain in the continuous casting crystallizer vibration system. The fault location of the transmission chain includes eccentric shaft phase deviation, connecting rod hinge clearance exceeding tolerance, and guide plate constraint slack.

[0070] Specifically, in step S385, the judgment results of steps S383 and S384 are combined to locate the fault location of the transmission chain based on the abnormal combination of directional pairs. Since faults in different links of the crystallizer's transmission chain will lead to different abnormal combinations of directional phases, the specific location of the fault can be inferred from the combination of directional pairs. For example, when only the phase difference of one directional pair is synchronously abnormal while the phase differences of the other two pairs are normal, it indicates that there is a fault in the transmission link involved in that directional pair, which may indicate a phase deviation in the eccentric shaft in that direction. When the phase differences of all three directional pairs are synchronously abnormal, it indicates that the fault occurs in the common transmission part of the three directions, which may indicate a phase deviation in the drive source. When the phase difference of one directional pair is jittery but the average value is normal, it indicates that there is a gap or constraint slack between the directional pairs, which may indicate that the linkage hinge gap in that direction is out of tolerance. When the phase differences of all three directional pairs are jittery, it indicates that the overall constraint stiffness of the system has decreased, which may indicate a constraint slack in the guide plate. The fault location method based on the above-mentioned direction pair combination can directly map the phase analysis results to specific fault locations in the transmission chain, providing maintenance personnel with accurate maintenance guidance.

[0071] Example 4 Based on the above embodiment 1, please refer to Figure 4 This embodiment provides another method for detecting fiber optic acceleration vibration in continuous casting crystallizers. The same content as in Embodiment 1 will not be repeated here; the difference lies in: In this embodiment, at least two fiber optic accelerometer probes are provided in each of the three orthogonal directions. The at least two fiber optic accelerometer probes in the same direction are installed at a predetermined distance along that direction. The acceleration signal set includes at least two acceleration signals corresponding to each direction. Before step S200, there are also the following steps S110-S130. It should be noted that steps S110-S130 are not specific steps of step S100. The sequence numbers of S110-S130 are only for the convenience of explanation.

[0072] S110: Verify the validity of at least two acceleration signals in the same direction and discard invalid signals.

[0073] Specifically, in step S110, the acceleration signals collected by at least two fiber optic accelerometer probes set in each orthogonal direction are individually validated to determine whether each signal truly and reliably reflects the vibration state in that direction. By eliminating invalid signals through validation, only valid signals that have passed all validations are retained for subsequent processing, thereby effectively avoiding the impact of measurement errors introduced by sensor malfunctions or external interference on the final detection results.

[0074] S120: When the number of valid signals in the same direction is two, calculate the composite acceleration signal in that direction based on the two valid signals; when the number of valid signals in the same direction is one or zero, perform fault tolerance processing.

[0075] Specifically, in step S120, different processing strategies are executed based on the number of remaining valid signals in the same direction after verification in step S110. When the number of valid signals in the same direction is two, both probes are in normal working condition, and the composite acceleration signal in that direction is calculated based on the two valid signals. When the number of valid signals in the same direction is one or zero, it indicates that the redundant measurement function in that direction has been partially or completely lost, and fault-tolerant processing is performed. Through the above-mentioned hierarchical processing mechanism, the system can still maintain basic detection functions when the probes partially fail, and can still provide reference data through prediction when all probes fail, effectively improving the fault tolerance and reliability of the detection system.

[0076] S130: Update the acceleration signal set using the synthetic acceleration signals or fault-tolerant processing results from each direction, and use the updated acceleration signal set for subsequent steps S200-S500.

[0077] Specifically, in step S130, the final signals obtained after processing in step S120 for each direction—including acceleration signals synthesized from dual effective signals, acceleration signals directly used from a single effective signal, or alternative signals extrapolated and predicted in the event of total failure—replace the original acceleration signals in the corresponding directions within the acceleration signal set, forming an updated acceleration signal set. The updated acceleration signal set eliminates interference from invalid signals and incorporates complementary information from redundant probes or fault-tolerant alternative data, resulting in higher reliability and quality compared to the original acquired acceleration signal set. Subsequently, the updated acceleration signal set is used as input, and the processing steps from S200 to S500 are executed sequentially, thereby ensuring the robustness and accuracy of the entire detection method under conditions of partial sensor failure or signal interference.

[0078] Further, in step S120, the composite acceleration signal in that direction is calculated based on the two valid signals, including the following steps S121-S125: S121: Obtain the installation spacing of two valid probes in the same direction along that direction.

[0079] Specifically, in step S121, the installation spacing between two effective fiber optic accelerometer probes in the same direction is obtained according to the pre-recorded installation parameters. This installation spacing is a fixed parameter determined during sensor installation and reflects the spatial positional difference between the two probes in the vibration propagation direction. Its magnitude directly affects the accuracy of subsequent calculations of vibration wave propagation velocity and vibration acceleration spatial gradient.

[0080] S122: Calculate the cross-correlation function between two valid signals to determine the time delay between the two signals.

[0081] Specifically, in step S122, two effective acceleration signals in the same direction are cross-correlated. The cross-correlation function describes the similarity of the signals at different time offsets. When the cross-correlation function value reaches its maximum at a certain time offset, that offset represents the time delay between the two signals. The physical meaning of this time delay is that when the vibration wave propagates along this direction in the crystallizer structure, it takes a certain amount of time to travel from the first probe position to the second probe position; this propagation time is the time delay. Accurately extracting the time delay through cross-correlation analysis provides a foundation for subsequent calculations of the vibration wave propagation velocity.

[0082] S123: Calculate the propagation speed of the vibration wave along this direction based on the installation spacing and time delay.

[0083] Specifically, in step S123, the installation spacing obtained in step S121 is divided by the time delay determined in step S122 to obtain the propagation speed of the vibration wave along that direction in the crystallizer structure. This propagation speed reflects the transmission efficiency of vibration energy along that direction and is a dynamic manifestation of the stiffness characteristics of the crystallizer structure. Differences in propagation speed in different directions can reflect the stiffness anisotropy of the structure in different directions. When the propagation speed in a certain direction changes significantly, it may indicate that the structural stiffness in that direction has changed.

[0084] S124: Weighted synthesis of two valid signals using the vibration wave propagation velocity as the weight:

[0085] in and These are the acceleration signals from the two effective probes, respectively. and These are the vibration wave propagation velocities calculated based on the positions of the two probes, when... It degenerates into an arithmetic mean; Simultaneously, based on the amplitude difference between the two valid signals and the installation spacing, the spatial gradient of vibration acceleration in this direction is calculated.

[0086] Specifically, in step S124, the vibration wave propagation velocity calculated in step S123 is used as a weighting coefficient to weight and synthesize the acceleration signals from the two valid probes. The probe signal with the larger propagation velocity is assigned a higher weight because a larger propagation velocity indicates a higher vibration transmission efficiency in that direction, and the signal measured by that probe is more representative of the true vibration state in that direction. Compared with the traditional arithmetic mean method, this propagation velocity-based weighted synthesis method fully considers the propagation characteristics of vibration waves in the structure, making the synthesis result more scientific and reasonable. Simultaneously, the spatial gradient of vibration acceleration in that direction is calculated by dividing the amplitude difference between the two valid signals at the same moment by the installation spacing. The spatial gradient of acceleration reflects the rate of change of vibration amplitude in space and is a quantitative indicator of the uniformity of the vibration field.

[0087] S125: When the spatial gradient of the vibration acceleration exceeds the preset gradient threshold, it is determined that there is an abnormality in the vibration field in that direction, and a corresponding warning message is generated.

[0088] Specifically, in step S125, the spatial gradient of vibration acceleration calculated in step S124 is compared with a preset gradient threshold. This gradient threshold can be determined based on the statistical distribution of spatial gradients measured multiple times during normal operation of the crystallizer. When the spatial gradient of vibration acceleration exceeds the preset gradient threshold, it indicates that the attenuation or amplification of the vibration amplitude in that direction with spatial position has exceeded the normal range, and it is determined that there is an abnormality in the vibration field in that direction. This abnormality may originate from local stiffness changes in the crystallizer structure in that direction, asymmetry of support constraints, or uneven distribution of transmission forces. After generating the corresponding warning information, this information can serve as a supplementary basis for subsequent anomaly judgment and type identification, or it can be directly output to the on-site operators as an equipment status prompt.

[0089] Furthermore, in step S110, the validity verification includes the following steps S111-S113: S111: Perform amplitude verification on each acceleration signal in the same direction: when the amplitude of the signal exceeds the preset effective amplitude range, the signal is determined to be invalid.

[0090] Specifically, in step S111, for each acceleration signal in the same direction, its amplitude is extracted and compared with a preset effective amplitude range. The effective amplitude range can be determined based on the maximum and minimum vibration amplitudes that the crystallizer may experience under the current operating parameters. The upper limit of this range should be higher than the maximum amplitude that may be reached during normal operation, and the lower limit should be higher than the noise level. When the signal amplitude exceeds this effective range, it indicates that the signal may have problems such as sensor saturation, abnormal signal transmission, or strong external impact interference, and cannot truly reflect the vibration state in that direction. The signal is then deemed invalid and removed from subsequent processing. Amplitude verification can effectively eliminate amplitude distortion signals caused by sensor damage or abnormal coupling.

[0091] S112: Perform frequency verification on each acceleration signal in the same direction: when the main frequency of the signal deviates from the current operating frequency of the continuous casting crystallizer by more than the preset frequency deviation threshold, the signal is determined to be invalid.

[0092] Specifically, in step S112, a fast Fourier transform is performed on each acceleration signal in the same direction to obtain its spectrum. The frequency corresponding to the maximum amplitude in the spectrum is extracted as the dominant frequency of the signal, and this dominant frequency is compared with the current operating frequency of the crystallizer. The current operating frequency is set by the crystallizer's drive system and is a known, definite value. When the dominant frequency of the signal deviates from the current operating frequency by more than a preset frequency deviation threshold, it indicates that the main frequency component of the signal does not originate from the normal vibration of the crystallizer, and may be affected by external vibration sources or abnormal oscillations in the sensor itself. The signal is then deemed invalid. Frequency verification can effectively eliminate abnormal frequency signals caused by external interference or resonant coupling.

[0093] S113: Perform self-consistency check on each acceleration signal in the same direction: calculate the autocorrelation function of the signal, and when a secondary peak exceeding 50% of the main frequency amplitude appears outside the main peak of the autocorrelation function, the signal is determined to contain abnormal harmonic components and marked as invalid.

[0094] Specifically, in step S113, the autocorrelation function of each acceleration signal in the same direction is calculated, and the peak distribution characteristics of the autocorrelation function are analyzed. For an ideal vibration signal, its autocorrelation function should not have significant secondary peaks outside the main peak. When a secondary peak exceeding a set proportion of the main peak amplitude appears at a certain position outside the main peak, it indicates that the signal contains non-intrinsic anomalous harmonic components, which may originate from problems such as nonlinear response of the probe, mechanical loosening, or resonant coupling. When such a situation occurs, the signal is determined to contain anomalous harmonic components, marked as invalid, and discarded. Self-consistency verification can effectively identify signals with severe waveform distortion and unreliability.

[0095] Furthermore, in step S120, the fault tolerance process includes the following steps S126-S127. It should be noted that the need for steps S126-S127 is only for illustrative purposes and is not intended to limit them to be located after step S125.

[0096] S126: When the number of valid signals in the same direction is one, and the probe corresponding to the valid signal is the only valid probe, the valid signal is used as the final signal in that direction and directly participates in subsequent processing, while generating a single probe working prompt message.

[0097] Specifically, in step S114, when only one of the two probes in the same direction passes the validity check, it indicates that the redundant measurement function in that direction has been partially lost. At this time, the acceleration signal from the only valid probe is directly used as the final signal for that direction, without further synthesis processing, ensuring that the direction can still provide measured signals for subsequent analysis. Simultaneously, a single-probe working prompt message is generated, including the identifier of that direction and the number of the failed probe, so that maintenance personnel can promptly arrange repairs to restore the redundant function.

[0098] S127: When the number of valid signals in the same direction is zero, extrapolate and predict the alternative signal at the current moment using the synthetic acceleration signal in that direction at the previous moment, and generate an alarm message indicating that all probes have failed.

[0099] Specifically, in step S115, when both probes in the same direction fail the validity check, it indicates that the measurement function in that direction has been completely lost, and the measured signal is unusable. At this time, extrapolation prediction is performed using the synthetic acceleration signal of that direction at the previous moment. Based on the periodicity and continuity of the vibration signal, the substitute signal for the current moment is estimated. Specifically, a linear extrapolation method based on vibration periodicity or a prediction method based on Kalman filtering can be used. Simultaneously, an alarm message indicating that all probes have failed is generated, including the identification of that direction and the failure status of both probes, to warn maintenance personnel that the measurement data in that direction is no longer the actual measured value and requires immediate action. Through the above fault-tolerant processing mechanism, it is ensured that the system can output a reference signal in that direction under any circumstances, guaranteeing the continuity of the detection process.

[0100] Example 5 Based on the above embodiment 1, please refer to Figure 5 This embodiment provides another method for detecting fiber optic acceleration vibration in continuous casting crystallizers. The same content as in Embodiment 1 will not be repeated here; the difference lies in: In this embodiment, at least two fiber optic accelerometers are provided in each of the three orthogonal directions, including a sensing fiber optic probe and a reference fiber optic probe. The first response signal output by the sensing fiber optic probe simultaneously includes wavelength drift components caused by vibration and temperature. The reference fiber optic probe uses the same fiber grating parameters as the sensing fiber optic probe and is installed in the same position, but the reference fiber optic probe is in an acceleration-unresponsive encapsulated state, and its second response signal only includes wavelength drift components caused by temperature. Before step S200, there are also steps S140-S170. It should be noted that the sequence numbers of steps S140-S170 are only for convenience of explanation and are not intended to be located after step S130.

[0101] S140: Synchronously acquire the first response signal of the sensing fiber optic probe and the second response signal of the reference fiber optic probe.

[0102] Specifically, in step S140, the response signals output by the sensing fiber optic probe and the reference fiber optic probe in the same direction are synchronously acquired by the signal processing unit. The first response signal of the sensing fiber optic probe reflects the center wavelength information under the combined effects of vibration and temperature, while the second response signal of the reference fiber optic probe only reflects the center wavelength information under the influence of temperature. The two probes are installed in the same location to ensure that they experience the same temperature environment. Synchronous acquisition ensures the temporal consistency of the two signals, providing an accurate temporal correspondence for subsequent differential compensation.

[0103] S150: Establish a temperature offset reference value under the current temperature conditions based on the second response signal.

[0104] Specifically, in step S150, the change in center wavelength of the reference fiber optic probe between the current time and the initial time is obtained. Since the reference fiber optic probe is in an acceleration-insensitive encapsulated state, this wavelength change is entirely caused by temperature changes. This wavelength change is used as the temperature offset reference value of the sensing fiber optic probe under the current temperature conditions. This reference value reflects in real time the degree of influence of ambient temperature on the fiber optic grating wavelength, providing a basis for subsequent elimination of the temperature component in the sensing fiber optic probe.

[0105] S160: The first response signal is differentially compensated in real time using the temperature offset reference value to obtain a vibration acceleration signal that eliminates the influence of temperature.

[0106] Specifically, in step S160, the change in center wavelength of the sensing fiber optic probe between the current moment and the initial moment is obtained. This change is caused by both vibration and temperature. The temperature offset reference value established in step S150 is subtracted from this total wavelength change to obtain the wavelength change caused purely by vibration. Then, based on the wavelength-strain sensitivity coefficient and acceleration-strain calibration coefficient of the fiber Bragg grating, the wavelength change caused purely by vibration is converted into a vibration acceleration value. Through the above differential compensation, the influence of temperature change on the measurement results of the sensing fiber optic probe is eliminated, resulting in a pure signal that only reflects vibration acceleration, effectively solving the problem of measurement drift caused by drastic temperature fluctuations in the continuous casting site.

[0107] S170: Update the acceleration signal set using the compensated and corrected vibration acceleration signal, and use the updated acceleration signal set for subsequent steps S200 to S500.

[0108] Specifically, in step S170, the vibration acceleration signal obtained after compensation and correction in step S160 replaces the original acceleration signal in the corresponding direction in the acceleration signal set, forming an updated acceleration signal set. The acceleration signal after temperature compensation and correction eliminates the measurement error introduced by changes in ambient temperature, and truly reflects the vibration acceleration of the crystallizer in that direction. Subsequently, the updated acceleration signal set is used as input, and the processing steps from S200 to S500 are executed sequentially, thereby ensuring the measurement accuracy and reliability of the entire detection method under the high-temperature and drastic changing environment of the continuous casting site.

[0109] Further, in step S160, the first response signal is subjected to real-time differential compensation using the temperature offset reference value, including the following steps S161-S165: S161: Obtain the first center wavelength of the sensing fiber optic probe. and the second center wavelength of the reference fiber optic probe .

[0110] Specifically, in step S161, the first center wavelength of the sensing fiber optic probe and the second center wavelength of the reference fiber optic probe at the current moment are synchronously read using a signal demodulator. These two center wavelength values ​​reflect the grating period changes of the sensing fiber optic probe under the combined effects of current temperature and vibration, and the grating period changes of the reference fiber optic probe under the current temperature, respectively. Synchronous reading ensures that the two signals are strictly corresponding in time, providing accurate raw data for subsequent differential compensation.

[0111] S162: Calculate the wavelength drift of the reference fiber optic probe between the current time and the initial time. , serving as a reference for temperature-induced wavelength shift.

[0112] Specifically, in step S162, the second center wavelength of the reference fiber optic probe at the current moment is subtracted from the second center wavelength at the initial moment from the second center wavelength of the reference fiber optic probe obtained in step S161 to obtain the wavelength drift of the reference fiber optic probe. Since the reference fiber optic probe is in an acceleration-insensitive encapsulated state, this wavelength drift is entirely caused by changes in ambient temperature. This wavelength drift is used as a reference for the temperature-induced wavelength shift under the current temperature conditions, representing the actual impact of temperature changes on the center wavelength of the fiber optic grating.

[0113] S163: Calculate the wavelength drift of the sensing fiber optic probe between the current time and the initial time. .

[0114] Specifically, in step S163, the wavelength drift of the fiber optic sensor is obtained by subtracting the initial center wavelength from the current center wavelength of the sensor probe acquired in step S161. The fiber optic sensor probe is sensitive to both vibration and temperature; therefore, this wavelength drift includes the wavelength change caused by both factors. Since the initial wavelength value is a known reference value, dynamic monitoring can be achieved by simply tracking the change in wavelength relative to the initial value.

[0115] S164: Subtract the wavelength drift reference caused by temperature from the total wavelength offset of the sensor's fiber optic probe to obtain the wavelength drift caused by pure vibration. .

[0116] Specifically, in step S164, the total wavelength drift of the sensing fiber optic probe obtained in step S163 is subtracted from the wavelength drift reference caused by temperature obtained in step S162 to obtain the wavelength drift caused by pure vibration. Through the above differential operation, the influence of temperature on wavelength drift is completely eliminated, and the remaining wavelength drift is caused only by vibration acceleration, thereby realizing real-time separation of vibration signal and temperature signal, effectively overcoming the problem of measurement drift caused by drastic temperature fluctuations in the continuous casting site.

[0117] S165: Based on the wavelength-strain sensitivity coefficient of fiber Bragg gratings and acceleration-strain calibration coefficient The wavelength shift caused by pure vibration is converted into vibration acceleration value. ; Meanwhile, real-time detection The rate of change, when When the rate of change exceeds the preset temperature change threshold, it is determined that the environment of the continuous casting crystallizer has undergone a sudden temperature change. At this time, the response time constant of temperature compensation is shortened to improve the tracking speed of compensation.

[0118] Specifically, in step S165, the wavelength drift caused by pure vibration obtained in step S164 is first multiplied by the wavelength-strain sensitivity coefficient of the fiber Bragg grating to obtain the strain value of the fiber Bragg grating; then, the strain value is multiplied by the acceleration-strain calibration coefficient to obtain the vibration acceleration value. Both coefficients are determined through calibration tests before the sensor leaves the factory. The wavelength-strain sensitivity coefficient reflects the wavelength drift caused by unit strain, and the acceleration-strain calibration coefficient reflects the strain caused by unit acceleration; both are inherent parameters of the sensor. Furthermore, the rate of change of the wavelength drift of the reference fiber optic probe is monitored in real time, reflecting the drastic change in the current ambient temperature. When this rate of change exceeds a preset temperature abrupt change threshold, it indicates a sudden temperature change in the environment of the continuous casting crystallizer. At this time, the system automatically shortens the response time constant of the temperature compensation. By increasing the tracking speed of compensation during rapid temperature changes and maintaining normal compensation smoothness during stable temperatures, adaptive adjustment of temperature compensation is achieved, enabling the system to maintain high-precision vibration measurement under the complex and variable temperature conditions of the continuous casting site.

[0119] The above description is merely a preferred embodiment of this application and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of the invention involved in this application is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the inventive concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features with similar functions disclosed in this application.

Claims

1. A method for detecting fiber optic acceleration vibration in a continuous casting crystallizer, characterized in that, The method includes: fiber optic acceleration probes are installed in three orthogonal directions of the continuous casting crystallizer; Acceleration signals are acquired by the fiber optic accelerometer probe to obtain an acceleration signal set, which includes at least one acceleration signal corresponding to each of the three orthogonal directions of the continuous casting crystallizer. The acceleration signal set is processed to obtain vibration information of the continuous casting crystallizer in three orthogonal directions, the vibration information including amplitude and frequency; Determine whether abnormal vibration has occurred based on the vibration information; When abnormal vibration occurs, the type of vibration abnormality is identified based on the vibration information. A detection report is generated based on the type of vibration anomaly.

2. The fiber optic acceleration vibration detection method according to claim 1, characterized in that, The step of determining whether abnormal vibration has occurred based on vibration information includes: The instantaneous amplitude sequence of vibration signals in three orthogonal directions is extracted, and the instantaneous amplitudes at each time point are connected in order of sampling time to construct a three-dimensional vibration phase trajectory curve. The curvature and deflection of the phase trajectory at each sampling point are calculated. Calculate the curvature deviation between the curvature and the theoretical curvature at each sampling point, and the deflection deviation between the deflection and the theoretical deflection. Also calculate the cumulative value of the curvature deviation and the cumulative value of the deflection within one vibration cycle. The theoretical curvature and theoretical deflection are determined by the theoretical vibration trajectory of the crystallizer under the current operating parameters. Abnormal vibration is defined as the cumulative value of curvature deviation exceeding the first threshold or the cumulative value of deflection deviation exceeding the second threshold. When both exceed the corresponding threshold, it is defined as severe abnormal vibration.

3. The fiber optic acceleration vibration detection method according to claim 2, characterized in that, The method of identifying vibration anomaly types based on vibration information includes: The time-domain and frequency-domain features of vibration signals in three orthogonal directions during the abnormal vibration period are extracted, and the distortion direction feature vector of the abnormal vibration in three-dimensional space is constructed based on the curvature deviation and deflection deviation. The time-domain features include at least peak value, root mean square value, kurtosis and waveform factor, and the frequency-domain features include at least the main frequency amplitude, harmonic energy ratio and sideband amplitude. The extracted time-domain features, frequency-domain features, and distortion direction feature vectors are input into a pre-established anomaly identification model, which outputs vibration anomaly types, including mechanical component wear anomalies, hydraulic system leakage anomalies, transmission mechanism failure anomalies, and misalignment anomalies.

4. The fiber optic acceleration vibration detection method according to claim 1, characterized in that, After determining whether abnormal vibration has occurred based on the vibration information, the method further includes: Calculate the distribution ratio of vibration energy in each direction at the fundamental frequency and each harmonic frequency to construct a three-dimensional energy distribution matrix; Calculate the norm deviation between the current energy distribution matrix and the reference energy distribution matrix under normal operating conditions; When the norm deviation exceeds a preset threshold, the spectral structure is determined to be abnormal.

5. The fiber optic acceleration vibration detection method according to claim 1, characterized in that, After determining whether abnormal vibration has occurred based on the vibration information, the method further includes: Calculate the instantaneous phase difference sequence between each pair of vibration signals in the three orthogonal directions, and calculate the mean and standard deviation of each phase difference sequence within one vibration cycle; When any mean value deviates from the theoretical phase difference value by more than the first deviation threshold, it is determined that there is a phase synchronization abnormality between the corresponding direction pairs. When any standard deviation exceeds the second deviation threshold, it is determined that there is a phase jitter abnormality between the corresponding direction pairs.

6. The fiber optic acceleration vibration detection method according to claim 1, characterized in that, At least two fiber optic accelerometers are provided in each of the three orthogonal directions. The at least two fiber optic accelerometers in the same direction are installed at a predetermined distance along that direction. The acceleration signal set includes at least two acceleration signals corresponding to each direction. Before processing the acceleration signal set, the method further includes: Verify the validity of at least two acceleration signals in the same direction and discard invalid signals; When there are two valid signals in the same direction, the composite acceleration signal in that direction is calculated based on the two valid signals; when there are one or zero valid signals in the same direction, fault tolerance processing is performed. The acceleration signal set is updated using the synthesized acceleration signals from each direction or the results of fault-tolerant processing.

7. The fiber optic acceleration vibration detection method according to claim 6, characterized in that, The calculation of the composite acceleration signal in this direction based on the two valid signals includes: The installation distance between two effective probes along this direction is obtained, the cross-correlation function between the two effective signals is calculated to determine the time delay between them, and the propagation speed of the vibration wave along this direction is calculated based on the installation distance and the time delay. The two valid signals are weighted and synthesized using the propagation speed as the weight, with the signal having a higher propagation speed having a higher weight. The spatial gradient of vibration acceleration in this direction is calculated based on the amplitude difference between the two valid signals and the installation spacing. When the spatial gradient exceeds a preset threshold, it is determined that there is an abnormality in the vibration field in this direction and an early warning is generated.

8. The fiber optic acceleration vibration detection method according to claim 6, characterized in that, The fault tolerance processing includes: When the number of valid signals in the same direction is one, and the probe corresponding to the valid signal is the only valid probe, the valid signal is used as the final signal in that direction and directly participates in subsequent processing, while generating a single probe working prompt message. When the number of valid signals in the same direction is zero, extrapolation prediction is performed using the synthetic acceleration signal in that direction at the previous moment to obtain the replacement signal at the current moment, and an alarm message indicating that all probes have failed is generated.

9. The fiber optic acceleration vibration detection method according to claim 1, characterized in that, At least two fiber optic accelerometers are provided in each of the three orthogonal directions, including a sensing fiber optic probe and a reference fiber optic probe. The first response signal output by the sensing fiber optic probe contains wavelength drift components caused by both vibration and temperature. The reference fiber optic probe uses the same fiber grating parameters as the sensing fiber optic probe and is installed in the same position. Its second response signal contains only the wavelength drift component caused by temperature. Before processing the acceleration signal set, the method further includes: Simultaneously acquire the first response signal and the second response signal, and establish a temperature offset reference value under the current temperature conditions based on the second response signal; The first response signal is differentially compensated in real time using the temperature offset reference value to obtain a vibration acceleration signal that eliminates the influence of temperature. The acceleration signal set is updated using the compensated and corrected vibration acceleration signal.

10. The fiber optic acceleration vibration detection method according to claim 9, characterized in that, The real-time differential compensation of the first response signal using the temperature offset reference value includes: Acquire the first wavelength drift of the sensing fiber optic probe between the current time and the initial time, and the second wavelength drift of the reference fiber optic probe between the current time and the initial time; Using the second wavelength drift as a reference for temperature-induced wavelength drift, the second wavelength drift is subtracted from the first wavelength drift to obtain the wavelength drift caused by pure vibration.

Images