Fiber grating-based valve multi-channel strain signal demodulation monitoring method

Abstract

This invention discloses a multi-channel strain signal demodulation and monitoring method for valves based on fiber Bragg gratings, belonging to the field of signal demodulation and monitoring technology. The method first deploys fiber Bragg grating sensors to form a sensing array at key parts of the valve. Combined with finite element transient dynamics simulation of the valve, a strain synergy map containing a structural synergy weight matrix and historical reference wavelength differences is constructed. Next, peak detection is performed on the reflection spectrum, and the spectral independent resolution and structural synergy deviation are calculated. Diagnostic quadrants are divided, and a two-factor diagnostic quadruple is output. Channels are split according to quadrants, and an objective function is constructed for indistinguishable channels to solve for the synergistic constraint demodulation wavelength. Finally, the normalized residual energy is calculated and refined in a directional manner, converting the demodulated wavelength into an engineering strain value output. This invention improves the accuracy and reliability of dense spectral channel demodulation, accurately reflecting the strain state of the valve structure and providing effective data support for valve health monitoring.

Description

Technical Field

[0001] This invention relates to the field of signal demodulation and monitoring technology, and more specifically, to a method for demodulating and monitoring multi-channel strain signals of valves based on fiber Bragg gratings. Background Technology

[0002] This invention belongs to the field of signal demodulation and monitoring technology, specifically involving the application of fiber optic grating (Fiber Bragg grating) sensing technology in the signal demodulation of valve structural strain monitoring. Fiber Bragg grating sensors are widely used in industrial valve structural strain monitoring due to their advantages such as resistance to electromagnetic interference, tolerance to harsh environments, and distributed monitoring capabilities. Strain quantification is achieved by detecting the center wavelength drift of the grating. A densely deployed multi-channel sensor can comprehensively capture the strain state of key parts of the valve, such as the valve body, valve stem, and sealing surface. The necessary multi-channel strain signal demodulation technology to achieve accurate spectral signal analysis is the core element for the successful implementation of fiber optic grating sensing systems in valve health monitoring.

[0003] Existing fiber Bragg grating multi-channel strain signal demodulation technologies for valves mostly employ independent peak detection and demodulation for each channel, failing to consider the structural synergy of strain in key valve components. When dense sensor deployment leads to overlapping spectral reflection peaks, independent demodulation is prone to peak positioning deviations, resulting in a significant decrease in spectral resolution. Furthermore, existing technologies lack a two-factor comprehensive diagnostic and constraint optimization mechanism for demodulation results, failing to effectively distinguish demodulation errors caused by spectral overlap from actual structural strain anomalies in the valve. Consequently, the reliability and accuracy of the demodulation results are insufficient, making it difficult to meet the high-precision, high-reliability structural strain monitoring requirements of industrial valves. To address these issues, this invention proposes a solution. Summary of the Invention

[0004] In order to overcome the above-mentioned defects of the prior art, embodiments of the present invention provide a valve multi-channel strain signal demodulation and monitoring method based on fiber optic gratings to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: The valve multi-channel strain signal demodulation monitoring method based on fiber optic grating includes the following steps: fiber optic grating sensors are deployed in key structural parts of the valve body to form a channel sensing array; and a valve structural strain synergy map containing the structural synergy weight matrix and the historical reference wavelength difference is constructed based on the transient dynamic simulation results of the valve finite element mechanical model. Independent peak detection is performed on each channel of the dense reflection spectrum of the current frame channel. The spectral independent resolution, which characterizes the resolvability of the reflection peak, is calculated from the detection results of each channel, and a reliable channel set is obtained accordingly. The structural coordination deviation, which characterizes the degree of deviation of each channel from the structural coordination prediction, is calculated by combining the measured wavelength in the reliable channel set with the strain coordination spectrum of the valve structure. Based on the spectral independent resolution and the structural coordination deviation, each channel is assigned to the corresponding diagnostic quadrant and a two-factor diagnostic comprehensive value is output. Based on the diagnostic quadrant markings, each channel is split. Channels with independently resolvable spectra directly adopt the measured wavelengths, while channels with indistinguishable spectra form a subset of constraint-solving channels. The comprehensive constraint strength is obtained by multiplying the spectral independent resolvability and the structural cooperative deviation by the modulation function and the constraint strength base coefficient. The objective function is constructed and solved using the product of the comprehensive constraint strength and the structural cooperative weight matrix as the constraint weights to obtain the cooperative constraint demodulation wavelength. The normalized residual energy at the demodulation wavelength of each channel is calculated. For channels with normalized residual energy exceeding the standard, targeted refinement is performed using the structural synergy weight matrix, and the final demodulation wavelength is converted into an engineering strain value output.

[0006] In a preferred embodiment, the structural synergy weight matrix is ​​constructed as follows: the covariance of the strain time series of any two channels under the same typical working condition is calculated and divided by the product of their respective standard deviations to obtain the dynamic correlation coefficient under the typical working condition. The maximum value of the absolute value of the dynamic correlation coefficient obtained under all typical working conditions is taken, and the maximum value is multiplied by the weight enhancement index to obtain the structural synergy weight. The weights of channels whose structural synergy weight values ​​are less than the graph edge weight truncation threshold are reset to zero.

[0007] In a preferred embodiment, the historical reference wavelength difference is calculated as follows: the strain time series of two channels in any channel pair under the same typical working condition are multiplied by their respective calibrated strain sensitivity coefficients and the difference is calculated. The average simulation time of this difference is taken to obtain the single working condition reference wavelength difference under that working condition. Then, the arithmetic mean of the single working condition reference wavelength differences under all typical working conditions is taken to obtain the historical reference wavelength difference.

[0008] In a preferred embodiment, the spectral independent resolution is calculated as follows: the smaller of the current wavelength spacing between the i-th channel and its left nearest neighbor channel and the current wavelength spacing between its right nearest neighbor channel is taken as the minimum neighbor channel spacing. The minimum neighbor channel spacing is divided by the product of the spectral resolution normalization coefficient and the average full width at half maximum (FWHM) to obtain the spacing component. The normalized peak intensity of the i-th channel is divided by the peak intensity determination benchmark to obtain the intensity component. The spacing component and the intensity component are multiplied to obtain the spectral independent resolution of the i-th channel. The missing side spacing of the leftmost and rightmost channels is taken as positive infinity. The normalized peak intensity is the ratio of the peak reflectance of the channel to the global maximum reflectance of the current frame, and i is the channel identifier.

[0009] In a preferred embodiment, the structural coordination deviation is calculated as follows: channels with spectral independent resolution not lower than the resolution threshold are included in the reliable channel set; the measured wavelength of each channel in the reliable channel set is added to the difference between its historical reference wavelength and that of the i-th channel to obtain the single-source prediction value of each reliable channel for the i-th channel; the structural coordination prediction wavelength of the i-th channel is obtained by weighting all single-source prediction values ​​with structural coordination weight; the structural coordination deviation of the i-th channel is obtained by dividing the absolute value of the difference between the spectral reference value of the i-th channel and the structural coordination prediction wavelength by the average full width at half maximum (FWHM); wherein the spectral reference value of the channel in the reliable channel set is the measured peak center wavelength obtained by fitting it, and the spectral reference value of the channel in the channel set to be constrained is its spectral indication wavelength.

[0010] In a preferred embodiment, the rule for classifying each channel into the corresponding diagnostic quadrant is as follows: when the spectral independent resolution is not lower than the resolution threshold and the structural coherence deviation does not exceed the deviation threshold, it is classified into quadrant I; when the spectral independent resolution is not lower than the resolution threshold and the structural coherence deviation exceeds the deviation threshold, it is classified into quadrant II; when the spectral independent resolution is lower than the resolution threshold and the structural coherence deviation does not exceed the deviation threshold, it is classified into quadrant III; and when the spectral independent resolution is lower than the resolution threshold and the structural coherence deviation exceeds the deviation threshold, it is classified into quadrant IV. The two-factor diagnostic composite value is calculated as follows: the quotient obtained by dividing the independent spectral resolution by the resolution threshold is limited to no more than 1, then 1 is subtracted and the square is taken. This square is then added to the square of the quotient obtained by dividing the structural co-deviation by the deviation threshold. The sum of the two is the two-factor diagnostic composite value.

[0011] In a preferred embodiment, the rules for the split processing are as follows: channels diagnosed as quadrant I directly use their measured peak center wavelength as the demodulated output value; channels diagnosed as quadrant II use their measured peak center wavelength as the demodulated output value and add a structural anomaly marker; channels diagnosed as quadrant III and quadrant IV form a subset of constraint-solving channels and enter the joint constraint optimization solution.

[0012] In a preferred embodiment, the comprehensive constraint strength is equal to the product of the constraint strength basic coefficient, the spectral factor modulation function value, and the structural factor modulation function value. The spectral factor modulation function is calculated by dividing the spectral independent resolvability by the resolvability determination threshold, limiting the quotient to no more than 1, subtracting 1 from the quotient, and then squaring the result to obtain the spectral factor modulation function value. This function value reflects the impact of the unreliability of spectral data on the constraint requirements. The structural factor modulation function is calculated by dividing the structural coordination deviation by the deviation judgment threshold, taking the square, adding 1 to the squared value, and taking the reciprocal to obtain the structural factor modulation function value. This function value reflects the influence of the reliability of structural coordination prediction on the constraint effectiveness.

[0013] In a preferred embodiment, the objective function comprises two parts: a data fidelity term and a structural co-constraint term. The data fidelity term is the sum of squares of the differences between the wavelength to be solved and the spectral indication wavelength of each channel in the constraint-solved channel subset. The structural co-constraint term is the weighted sum of the squares of the wavelength differences between each channel in the constraint-solved channel subset and the remaining channels among all N channels, deviating from the corresponding historical reference wavelength. The weight is the product of the comprehensive constraint strength of the channel and the structural co-constraint weight of the corresponding channel pair. Channels belonging to the constraint-solved channel subset are represented by the wavelength to be solved, while channels not belonging to the constraint-solved channel subset are represented by the determined demodulation output value. The partial derivatives of each wavelength to be solved in the objective function are calculated and set to zero to obtain a system of K linear equations, where K is the number of channels in the constraint-solved channel subset. Solving this system of linear equations yields the co-constrained demodulation wavelength.

[0014] In a preferred embodiment, the targeted refinement is performed as follows: based on the demodulation wavelength of each channel and the obtained normalized peak intensity and average full width at half maximum (FWHM) of each channel, the theoretical reflection spectrum is reconstructed using a Gaussian superposition model. The absolute value of the difference between the measured spectrum and the reconstructed spectrum at the demodulation wavelength of each channel is calculated and divided by the global maximum reflection intensity of the current frame to obtain the normalized residual energy. For channels whose normalized residual energy exceeds the residual energy judgment threshold, the refinement step size coefficient is multiplied by the ratio of the normalized residual energy of that channel to the global maximum reflection intensity of the current frame, and then multiplied by the weighted sum of the wavelength difference between that channel and all other channels deviating from the corresponding historical reference wavelength, with structural synergy weight as the weight, to obtain the correction amount. The correction amount is superimposed on the demodulation wavelength to obtain the refined wavelength. If the normalized residual energy recalculated after refinement is greater than the normalized residual energy before refinement, the refinement is canceled and the original demodulation wavelength is restored.

[0015] The technical effects and advantages of this invention's valve multi-channel strain signal demodulation monitoring method based on fiber Bragg grating are as follows: It constructs a valve structural strain coordination map, utilizing structural coordination information to provide demodulation constraints for spectral overlap channels, effectively solving the demodulation problem caused by overlapping spectral peaks in dense sensor arrays. This significantly improves the accuracy and reliability of multi-channel strain signal demodulation. A dual-factor diagnostic system of spectral independent resolvability and structural coordination deviation is proposed, enabling accurate classification and quantitative evaluation of the demodulation state of each channel. This effectively distinguishes between spectral demodulation errors and actual valve structural strain anomalies. Simultaneously, differentiated channel processing is achieved through diagnostic quadrants, ensuring demodulation efficiency for resolvable channels while optimizing non-resolvable channels through targeted coordination constraints, enhancing the intelligence and adaptability of the demodulation system. This is further improved by reconstructing the theoretical reflectance spectrum. To verify the rationality of the demodulation wavelength, the deviation channel is directionally corrected based on structural synergy weights, and a fine-tuning backoff rule is set to further eliminate demodulation deviations, ensuring that the demodulation results highly match the actual spectrum and valve structural strain characteristics. The final output demodulation result set contains multi-dimensional information such as strain values, diagnostic indicators, and anomaly markers, providing a comprehensive and accurate data source for valve structural health status assessment and trend analysis. It is suitable for the actual monitoring needs of industrial valves. The structural synergy map is calculated and solidified offline only during the initialization stage. Subsequent demodulation only requires real-time execution of spectral detection, constraint solving, and fine-tuning steps, balancing demodulation accuracy and real-time performance. Moreover, each preset parameter can be flexibly adjusted according to the valve structure and monitoring accuracy requirements, exhibiting good versatility and engineering practicality, and can be adapted to strain monitoring of industrial valves of different models and under different operating conditions. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating the valve multi-channel strain signal demodulation and monitoring method based on fiber Bragg grating according to the present invention. Figure 2This is a schematic diagram of the data relationship in the valve multi-channel strain signal demodulation and monitoring method based on fiber optic gratings according to the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Example Please see Figure 1 As shown, this invention discloses a method for demodulating and monitoring multi-channel strain signals of valves based on fiber Bragg gratings, comprising the following steps: Please see Figure 2 Step 1: Fiber grating sensors are deployed in key structural parts of the valve body to form a channel sensing array. Based on the transient dynamic simulation results of the valve finite element mechanical model, a valve structural strain synergy map containing the structural synergy weight matrix and the historical reference wavelength difference is constructed. Based on the valve's structure and monitoring objectives, fiber Bragg grating sensors are deployed at N key structural locations on the valve body. Each fiber Bragg grating sensor corresponds to a unique channel i to channel N, with channel i corresponding to the i-th key structural monitoring location on the valve, where i is the channel identifier, forming an N-channel sensor array. The exemplary sensors are fixed using high-temperature resistant epoxy adhesive with a temperature range of -40℃ to 200℃. Before fixing, the valve mounting surface is sandblasted to remove rust and cleaned with anhydrous ethanol. The sensor grating segment is tightly fitted to the valve surface, with the length direction of the grating segment aligned with the principal strain direction of the channel. The fiber optic leads are protected by corrugated tubing and run along the valve body, ultimately forming N channels. The deployment locations of the fiber Bragg grating sensors include, but are not limited to, the valve body's pressure-bearing wall surface, the circumferential section at the valve stem root, the sealing surface transition area, and the flange connection area. Each fiber Bragg grating sensor has its own unique initial center wavelength. ,in All initial center wavelengths are arranged sequentially within the operating band of the light source, and the initial wavelength spacing between adjacent channels is set according to the expected strain range of each channel; in this embodiment, the initial value is determined based on the key structural parts corresponding to the valve model. Take 16; Secondly, a finite element mechanical model consistent with the actual valve is established, and a selection is made. Transient dynamic simulation calculations were performed on representative typical operating conditions of various valves; the number of such typical operating conditions... These are preset parameters, with an initial value of 3, corresponding to the fully open transient condition, the fully closed transient condition, and the fluid impact condition, respectively; the simulation calculation time for each condition is [duration missing]. , The initial value is 10 seconds; After the simulation is completed, extract each working condition from the finite element results. Below, all Strain time series at each channel location ; where superscript Indicates the first Various working conditions and , The time variable is used for simulation; the strain time series is sampled at equal time intervals, and the sampling interval is consistent with the sampling period of the subsequent actual demodulation. It should be noted that this refers to analyzing the dynamic response process of the structure under time-varying loads. In this invention, it is used to calculate the strain variation of each channel of the valve under different operating conditions over time. For example, the typical simulation operating condition, namely the fully open transient condition, simulates the mechanical state of the valve stem bearing tensile load and the valve body bearing sudden pressure load in the flow channel during the rapid switching of the valve from the fully closed position to the fully open position. The fully closed transient condition simulates the mechanical state of the valve stem bearing compressive load and the sealing surface bearing contact extrusion load during the rapid switching of the valve from the fully open position to the fully closed position. The fluid impact condition simulates the mechanical state of the valve body wall bearing shear load when water hammer or sudden pressure change occurs in the pipeline. For each operating condition, transient mechanical simulation is performed, extracting the strain time series of all N channels under that condition. The number of sampling points for each time series is not less than 500, and the sampling time window should cover the complete transition process from the start to the steady state of that condition. To quantize any two channels and To assess the consistency of strain change trends under the same operating conditions, a dynamic correlation coefficient is calculated for the strain time series of the two channels. The dynamic correlation coefficient characterizes the degree of linear correlation between the strain response signals of the two channels under specific operating conditions and their changes over time. The calculation method is as follows: ;in, The covariance operation over the entire simulation time series is implemented in this invention as follows: Let the first... The simulation time series under various working conditions was obtained by sampling at equal time intervals. There are discrete sampling points, and the sampling time is... ,aisle The discrete values ​​of strain ,aisle The discrete values ​​of strain The covariance operation is specifically as follows: , In the formula , Channels , In the The mean of the strain time series under various working conditions; and The first and the The channel is in Standard deviation of strain sequence under various working conditions; The range of values ​​is The closer the absolute value is to 1, the more synergistic the strain changes of the two channels are; that is, positive values ​​are synergistic in the same direction and negative values ​​are synergistic in opposite directions. Considering that the same pair of channels may exhibit different cooperative characteristics under different operating conditions, for example, adjacent channels in the circumferential direction of the valve stem may be highly positively correlated under the fully closed condition due to axial preload, but exhibit different correlation patterns under the fluid impact condition due to lateral bending. Therefore, it is necessary to fuse the correlation coefficients under multiple operating conditions. This invention takes the maximum absolute value of the correlation coefficient under each operating condition as the fusion result, and expands the weight difference between high and low cooperative channel pairs through power-law enhancement operations, thereby constructing a structural cooperative weight matrix. The specific calculation method is as follows: ;in, The weight enhancement index is initially set to 3. Extensive simulations have verified that the cubic calculation can increase the weight difference between high-coherence and low-coherence channel pairs from a factor of 3 to a factor of 27. This effectively highlights the correlation strength of highly coherent channel pairs without causing weight polarization due to excessively high exponentiation. For example, the weights obtained for channel pairs with an original correlation coefficient absolute value of 0.9 are... The channel pairs with a correlation coefficient of 0.3 have only received weights of [missing value]. The difference between the two increased from 3 times to 27 times, thus effectively highlighting the correlation strength between channel pairs with strong structural synergy; To eliminate computational interference introduced by physically weakly correlated channels, the constructed structural synergy weight matrix is ​​truncated: when the structural synergy weight matrix... The calculated value is less than the graph edge weight truncation threshold. When this happens, the weight value is set to zero; graph edge weight truncation threshold. The critical weight value used to determine whether the structural synergy between channel pairs has engineering significance is the minimum weight criterion for determining effective associated edges in the valve structure strain synergy map. Its initial value is 0.05, which can be adjusted according to the valve structure. When the valve structure is complex and the channel synergy differences are large, a smaller value, such as 0.01~0.03, is used to retain more synergistic relationships; when the valve structure is simple and the channel synergy differences are small, a larger value, such as 0.07~0.1, is used to remove weakly associated channel pairs. After truncation, only the non-zero weights between channel pairs with significant structural synergy are retained in the structural synergy weight matrix, forming the valve structural strain synergy map. Valid associated edges; Using the strain data from the finite element simulation, combined with the strain sensitivity coefficients obtained from the calibration of each fiber Bragg grating sensor... Calculate any channel pair under various operating conditions The wavelength difference caused by strain between them, and its effect on The historical reference wavelength difference is obtained by taking the time average and the operating condition average for each operating condition. The specific calculation method is as follows: ;in, Indicates the simulation time Take the average; For the channel under typical working conditions With channel The difference in expected wavelength caused by structural strain differences; The final output of this step is defined as the valve structure strain synergy map. It includes the following three elements: Each channel node, and the truncated structural collaborative weight matrix. and historical reference wavelength difference matrix The The offline calculation is completed once during the demodulation initialization stage and stored in a fixed manner. In the subsequent real-time demodulation process of each frame of spectral data, it is repeatedly called as a fixed prior parameter in steps two, three and four.

[0019] Step 2: Perform independent peak detection on each channel of the dense reflection spectrum of the current frame. Calculate the spectral independent resolution, which characterizes the resolvability of the reflection peaks, based on the detection results of each channel. Based on this, a reliable channel set is obtained. Calculate the structural coordination deviation, which characterizes the degree of deviation of each channel from the structural coordination prediction, using the measured wavelengths in the reliable channel set combined with the valve structure strain coordination spectrum. Based on the spectral independent resolution and the structural coordination deviation, classify each channel into the corresponding diagnostic quadrant and output a two-factor diagnostic comprehensive value. This step involves analyzing the N-channel dense reflectance spectrum acquired in the current frame. First, independent peak detection is performed channel by channel; specifically, the demodulation wavelength of the previous frame for each channel is used, with the initial center wavelength used in the first frame. The search center is set with a preset search radius on each side. The spectral region within the range is fitted with a Gaussian function using the Levenberg-Marquardt algorithm; a preset search radius is used. The initial value is taken ,in The average full width at half maximum (FWHM) of the fiber grating reflection peaks in all channels, and the full WHM of the Gaussian reflection peaks. with Gaussian standard deviation Satisfies classic conversion relationships: Therefore, the mean Gaussian standard deviation parameter ,in Use fixed conversion factors; For channels that converge during the fitting process, record the center wavelength of the peak obtained from the fitting. and normalized peak intensity The peak reflectance and the global maximum reflectance of the spectrum are given. The ratio, the normalized peak intensity, takes values ​​in the range (0,1]. This indicates that the channel has extremely high reflection peak intensity and strong spectral distinctiveness; It indicates moderate intensity and average recognizability; This indicates a low intensity channel that is easily blocked by adjacent channels. For channels that do not meet the convergence criteria, the centroid position of the spectral energy distribution within the search region is recorded as the spectral cue wavelength. ; The convergence criterion is the termination criterion for the Gaussian function fitting iteration process, which includes two core parameters: the fitting parameter iteration deviation threshold and the fitting residual sum of squares threshold. During the fitting process, if the peak center wavelength deviation between two consecutive iterations is less than or equal to the fitting parameter iteration deviation threshold and the fitting residual sum of squares threshold is less than or equal to the fitting residual sum of squares threshold, the fitting is considered converged. The initial value of the fitting parameter iteration deviation threshold is 0.1 pm, and the initial value of the fitting residual sum of squares threshold is... Both can be dynamically adjusted according to the demodulation accuracy requirements; To quantitatively evaluate the first from the perspective of spectral data The degree to which a channel's reflection peak is obscured by adjacent channels in the current frame defines spectral independent resolvability. This indicator takes into account the first... The wavelength spacing between a channel and its left and right nearest neighbor channels, as well as the distinguishability of the intensity of the channel's own reflection peak, are calculated as follows: ; in, and The first The distance between the current wavelength position of a channel and the current wavelength positions of its left and right nearest neighbors, and for the leftmost and rightmost channels, the missing side distance is taken as... ; Let be the spectral resolution normalization coefficient, initially set to 1.0, such that when the nearest neighbor distance is exactly equal to one... hour, The spacing component is 1.0; The peak intensity is used as the benchmark for determining peak intensity. Its initial value is 0.5, which means that channels with peak intensity lower than half of the global maximum value will have reduced resolution due to intensity factors. Spectral independent resolution The larger the value, the easier it is for the reflection peak of that channel to be independently resolved in the spectrum; spectral independent resolution. The smaller the value, the more severely the channel is submerged in the overlapping area of ​​adjacent channels; The obtained spectral independent resolution Compared with the preset distinguishability threshold The reliability of channel demodulation is compared and classified, with a preset resolution threshold. This is the core critical value for classifying channel demodulation reliability, representing the lowest effective engineering criterion for spectral independent resolution, used to quantitatively distinguish reliable channels from channels to be constrained. The initial value is 0.5, and it is dynamically adjusted according to the accuracy requirements of the actual monitoring scenario. when At that time, it was considered that the reflection peak of this channel possessed independent resolution engineering validity, and the demodulation results had basic data credibility, thus it was included in the reliable channel set. ;when At that time, it was considered that the reflection peak of this channel was significantly affected by the overlap of adjacent channels, and the reliability of the independent demodulation results was insufficient, so it needed to be included in the set of channels to be constrained. Demodulation is aided by structural coordination information; To quantitatively evaluate the first from a structural physics perspective The degree of agreement between the current spectral information of the channel and the cooperative relationship between the valve structure and the data requires first utilizing a reliable set of channels. The measured wavelength of the channel, combined with the structural synergy weight matrix obtained in step one. Difference from historical reference wavelength The weighted prediction is used to calculate the first... The wavelength position that a channel should theoretically appear in the current frame if structural coherence is satisfied, i.e., the structural coherence predicted wavelength. The calculation method is as follows: In the formula, This indicates that the summation range includes all reliable channels, and j≠i excludes channel i itself. The structural synergy weights between channel i and reliable channel j; For reliable channels The measured wavelength; The physical meaning of this formula is: each reliable channel Measured wavelength Add channel and Historical reference wavelength difference Obtain the channel For the channel The single-source predicted value is then weighted with structural synergy. A weighted average is performed to obtain the comprehensive predicted wavelength; the channel with the larger weight contributes more to the prediction result, that is, structurally, it is more related to the channel. The stronger the synergy of a channel, the greater its predictive power. Obtaining structural co-predicted wavelengths Afterwards, structural coordination deviation Spectral reference value Different values ​​are used depending on the set affiliation of channel i. Channel i in the spectral reference value Take the measured peak center wavelength obtained from the fitting. ,for Channel i in the spectral reference value Hint wavelength for spectral energy centroid calculation Define structural coordination deviation as follows: ;in For channel The measured or indicated wavelength of the spectrum is the reliable channel. Unconstrained channel selection Structural Coordination Deviation The smaller the value, the higher the consistency between the spectral information of this channel and the structural co-prediction; structural co-prediction deviation. The larger the value, the more the spectral information of the channel deviates significantly from the structural co-prediction, which may mean that the channel has spectral overlap leading to mislocation, or that the valve in this region has indeed experienced an abnormal strain state that deviates from typical operating conditions. Based on spectral independent resolution and structural coordination deviation Threshold is determined by distinguishability. and deviation determination threshold To define the boundaries, each channel is assigned to one of the following four diagnostic quadrants: when and If it is classified as Quadrant I, it means that the spectrum can be independently resolved and is consistent with the structural prediction. The demodulation result of this channel can be directly accepted. when and It is then classified as quadrant II, indicating that the spectrum can be independently resolved, but deviates significantly from the structural co-prediction. The demodulation result of this channel is retained but marked as a candidate for structural anomaly. when and It is then classified as quadrant III, indicating that the spectrum cannot be independently resolved, but the spectral indications are basically consistent with the structural predictions. Cooperative constraint demodulation needs to be initiated and the structural constraints have high reliability. when and It is then classified as quadrant IV, where the spectrum cannot be independently resolved and the spectrum indicates a deviation from the structural prediction. Cooperative constraint demodulation needs to be initiated, but the constraint confidence needs to be reduced. To connect the above discrete classification with continuous quantitative analysis, a two-factor diagnostic composite value is defined. The calculation method is as follows: In the formula, the comprehensive value of the two-factor diagnosis is... In a two-dimensional diagnostic plane comprised of normalized spectral resolution and normalized structural deviation, the distance from the current channel state point to the ideal operating point is... , Distance measurement at time; A larger value indicates that the channel is more difficult to demodulate; The final output of this step is a two-factor diagnostic quadruple for each channel. ,in This serves as the diagnostic quadrant identifier; the quadruple is used as a whole and is called channel by channel in step three to control the selection of processing paths and the calculation of constraint strength, respectively.

[0020] Step 3: Based on the diagnostic quadrant markings, each channel is split. Channels with independently resolvable spectra directly adopt the measured wavelength. Channels with indistinguishable spectra form a subset of constraint solution channels. The spectral independent resolvability and the structural cooperative deviation are respectively processed by the modulation function and multiplied by the basic coefficient of constraint strength to obtain the comprehensive constraint strength. The product of the comprehensive constraint strength and the structural cooperative weight matrix is ​​used as the constraint weight to construct and solve the objective function to obtain the cooperative constraint demodulation wavelength. The channel splitting processing rules for this step are as follows: Based on the quadrant labels output in step two Perform the following traffic splitting process on N channels: For channels diagnosed as quadrant I, their independently demodulated wavelengths are... Simultaneously satisfying both spectral resolvability and structural consistency, therefore directly... As the demodulation output value of this channel No further constraint optimization is needed; For channels diagnosed as quadrant II, although their independently demodulated wavelengths are reliable based on spectral data, they significantly deviate from structural co-prediction, therefore they will also be... As the demodulation output value of this channel At the same time, structural anomaly markers are added to the output. Quadrant I channel This is to alert the subsequent monitoring system to potential local abnormal strains in the channel; For channels diagnosed as quadrants III and IV, since their spectra cannot be independently resolved, constraint-assisted demodulation must be performed using valve structure synergy information; all channels in quadrants III and IV are combined into a constraint-solving channel subset. ,right The following joint constraint optimization is performed on the channels within: Constraint solution channel subset Each channel in Its constraint strength Spectral independent resolution output from step two and structural coordination deviation Simultaneous regulation; specifically, defining modulation functions corresponding to the two parameters respectively: First, the spectral factor modulation function The set of data reflecting the impact of the unreliability of spectral data on constraint requirements The smaller the value, the greater the constraint requirement; the function is defined as: ; when At that time, it was completely indistinguishable. This indicates that the maximum constraint force is required; when hour, , indicating no constraints are required; the quadratic form ensures that the modulation function is within the bounds of the expression. near Smooth transition; Second, the structure factor modulation function The impact of the reliability of structural collaborative predictions on constraint effectiveness, i.e. The smaller the value, and the more consistent the spectral indications are with the structural predictions, the more reliable the structural constraints are, and the greater the constraint effectiveness should be assigned. The larger the value, the lower the reliability of the structural constraints, and the corresponding decrease in constraint effectiveness should occur; the function is defined as: ; when At the same time, completely consistent. The binding force is greatest when... hour, The binding force is reduced to half; when hour, When the value approaches zero, the binding force almost disappears. After obtaining the two modulation functions mentioned above, the first... Comprehensive constraint strength of the channel It is determined by the product of the two factors and the base coefficient, and the specific calculation method is as follows: ;in, The basic coefficient for constraint strength is initially set to 2.0; the comprehensive constraint strength... The computational logic is that a significant structural constraint is applied to a channel only when it simultaneously satisfies two conditions: it is not independently distinguishable on the spectrum and the structural co-prediction is reliable. If the spectrum is distinguishable, no constraint is applied; if the structural prediction is unreliable, the constraint is weakened. This dual-factor joint regulation mechanism ensures that the constraint intensity is always within a physically reasonable range, so that neither excessive constraint will mask the true strain anomaly, nor insufficient constraint will cause the demodulation failure of overlapping channels. Obtaining a subset of constraint solution channels Comprehensive constraint strength of each channel Then, construct the following objective function. It consists of two parts: data fidelity terms and structural coordination constraints. The first item is the data fidelity item, which requires the demodulation wavelength. It should not deviate excessively from the wavelength indicated by the spectrum. The first term preserves the information contribution of the spectral data itself; the second term is a structural cooperative constraint term, which requires that any two channels... and wavelength difference It should be close to the historical reference wavelength difference. ,in exist When taking the variable to be solved ,exist Take the determined output value at time That is, the direct acceptance results of quadrants I and II; the weights of the constraint terms are determined by... and The product of these two factors jointly controls the outcome; the former comes from two-factor diagnosis, and the latter comes from structural synergy maps. Neither can be dispensed with. The objective function For about By taking the partial derivatives of a quadratic function and setting them all to zero, we can obtain a system of linear equations; let... The CCP includes One channel and Then it forms Linear equation system of order ,in for A symmetric positive definite coefficient matrix. for A constant vector of dimension, all derived from known quantities ( , , , , The system of linear equations is constructed by directly solving the system and... Usually not exceeding 10, to obtain Cooperatively constrained demodulation wavelengths of all channels ; Finally, the direct acceptance results of quadrants I and II are combined with the cooperative constraint demodulation results of quadrants III and IV to form the complete set of data for this frame. Demodulation wavelength set of each channel Together with the diagnostic quadrant labels of each channel and exception markers This information is then transferred to step four for residual refinement and strain output.

[0021] Step 4: Calculate the normalized residual energy at the demodulation wavelength of each channel. For channels with normalized residual energy exceeding the standard, perform directional refinement using the structural synergy weight matrix, and convert the final demodulation wavelength into an engineering strain value output. This step utilizes the demodulation wavelengths of all channels output from step three. and the normalized peak intensity of each channel obtained in step two. and average full width at half maximum The theoretical reflectance spectrum is reconstructed according to the Gaussian superposition model; the reconstructed spectrum At any wavelength The calculation method at this location is as follows: ;in The Gaussian standard deviation parameter is calculated from the average full width at half maximum (FWHM). This represents the global maximum reflectance of the measured spectrum in the current frame. Subsequently, the demodulation wavelength for each channel... Calculate the measured spectrum at the location. With reconstructed spectrum Normalized residual energy between The specific calculation method is as follows: ;in The range of values ​​is A larger value indicates a worse match between the demodulated wavelength output in step three and the actual spectrum at that channel, suggesting a possible slight positioning error. Let the residual energy threshold be... , The initial value is 0.1; when When the demodulation wavelength of the channel meets the accuracy requirements, no correction is made; when At that time, it was determined that the passage needed further refinement; for The channel utilizes the structural synergy weights established in step one. Difference from historical reference wavelength The demodulation wavelength is then refined in a directional manner. The physical basis for this directional refinement is that if a channel's demodulation wavelength has a slight deviation, this deviation should have a consistent direction of manifestation in its structurally cooperating neighbor channels. Therefore, the refinement direction is determined by the weighted average of the cooperative prediction deviations of its neighbor channels, and the refinement step size is jointly controlled by the residual energy and the overall diagnostic state. The amount of correction required for each channel needs fine-tuning. The calculation method is as follows: ;in, To refine the step size coefficient, its initial value is set to 0.15; the items in parentheses... The physical meaning is: the current channel With channel The deviation of the wavelength difference from the historical reference value; if this deviation is positive, it indicates that the channel... The wavelength may be too low or the channel Too high, the correction direction should be... Increase; with After weighted summation, neighbor channels with strong structural synergy have a dominant contribution to the correction direction; The final demodulation wavelength after refinement is ; for The passage directly caused No corrections will be made. To prevent the fine-tuning process from introducing new deviations, the residual energy is recalculated for the fine-tuned wavelength. ;like If the residual error increases after refinement, then the refinement should be cancelled and the original refinement should be restored. This rollback mechanism ensures the monotony of the refinement process, meaning that refinement can only improve the result, not worsen it. After obtaining all The final demodulation wavelength of each channel Then, the wavelength shift of each channel is converted into engineering strain values. The specific calculation method is as follows: ;in For the first The initial center wavelength of the channel fiber grating. This is the strain sensitivity coefficient obtained by calibration of this channel, in units of... ; To verify the overall rationality of the demodulation results, a multi-channel consistency check was performed: (This was done on the data from step one.) All channels Calculate the deviation between its strain difference value and the historical reference difference value. If all channels are paired All within the allowable deviation of consistency in, The initial value is taken If the demodulation result of this frame passes the consistency check, then the demodulation result of this frame is determined to have passed the consistency check; if a channel pair exists... If the channel is not found to be in the required channel, then the corresponding channel will be marked as the channel that needs to be reviewed, and a review identifier will be added to the output result. The final output of this step is the complete demodulation result set of the current frame, which includes the following: All Final strain values ​​of each channel Two-factor diagnostic quadrant labels for each channel Structural anomaly markers Residual energy The output result set fully describes the strain state of each channel of the valve at the current moment and its demodulation quality assessment, and can be directly used for valve structural health status assessment and trend analysis. Steps one through four are executed sequentially when each frame of spectral data arrives. The output of step one is a fixed parameter, which is executed only once during initialization. Steps two through four are executed in real time frame by frame, forming a complete processing cycle for the valve multi-channel strain signal demodulation monitoring method based on fiber Bragg grating.

[0022] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0023] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0024] Those skilled in the art will recognize that the modules and algorithm modules of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and inventive constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0025] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0026] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0027] In conclusion, the above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for demodulating and monitoring multi-channel strain signals of valves based on fiber Bragg gratings, characterized in that, Includes the following steps: Fiber grating sensors are deployed in key structural parts of the valve body to form a channel sensing array. Based on the transient dynamic simulation results of the valve finite element mechanical model, a valve structural strain synergy map containing the structural synergy weight matrix and the historical reference wavelength difference is constructed. Independent peak detection is performed on each channel of the dense reflection spectrum of the current frame channel. The spectral independent resolution, which characterizes the resolvability of the reflection peak, is calculated from the detection results of each channel, and a reliable channel set is obtained accordingly. The structural coordination deviation, which characterizes the degree of deviation of each channel from the structural coordination prediction, is calculated by combining the measured wavelength in the reliable channel set with the strain coordination spectrum of the valve structure. Based on the spectral independent resolution and the structural coordination deviation, each channel is assigned to the corresponding diagnostic quadrant and a two-factor diagnostic comprehensive value is output. Based on the diagnostic quadrant markings, each channel is split. Channels with independently resolvable spectra directly adopt the measured wavelengths, while channels with indistinguishable spectra form a subset of constraint-solving channels. The comprehensive constraint strength is obtained by multiplying the spectral independent resolvability and the structural cooperative deviation by the basic coefficient of constraint strength after modulation function calculation. The objective function is constructed and solved using the product of the comprehensive constraint strength and the structural cooperative weight matrix as the constraint weights to obtain the cooperative constraint demodulation wavelength. The normalized residual energy at the demodulation wavelength of each channel is calculated. For channels with normalized residual energy exceeding the standard, targeted refinement is performed using the structural synergy weight matrix, and the final demodulation wavelength is converted into an engineering strain value output.

2. The method for demodulating and monitoring multi-channel strain signals of valves based on fiber Bragg gratings according to claim 1, characterized in that, The structural synergy weight matrix is ​​constructed as follows: calculate the covariance of the strain time series of any two channels under the same typical working condition and divide it by the product of their respective standard deviations to obtain the dynamic correlation coefficient under the typical working condition. Take the maximum value among the absolute values ​​of the dynamic correlation coefficients obtained under all typical working conditions, and perform a power operation on the maximum value with the weight enhancement index as the exponent to obtain the structural synergy weight. Set the weights of channels whose structural synergy weight values ​​are less than the graph edge weight truncation threshold to zero.

3. The method for demodulating and monitoring multi-channel strain signals of valves based on fiber Bragg gratings according to claim 1, characterized in that, The historical reference wavelength difference is calculated as follows: for any channel pair, the strain time series of the two channels under the same typical working condition are multiplied by their respective calibrated strain sensitivity coefficients and the difference is calculated. The average simulation time of this difference is taken to obtain the single working condition reference wavelength difference under that working condition. Then, the arithmetic mean of the single working condition reference wavelength differences under all typical working conditions is taken to obtain the historical reference wavelength difference.

4. The method for demodulating and monitoring multi-channel strain signals of valves based on fiber Bragg gratings according to claim 1, characterized in that, The spectral independent resolution is calculated as follows: the smaller of the current wavelength distance between the i-th channel and its left nearest neighbor channel and the current wavelength distance between the i-th channel and its right nearest neighbor channel is taken as the minimum neighbor channel distance. The minimum neighbor channel distance is divided by the product of the spectral resolution normalization coefficient and the average full width at half maximum (FWHM) to obtain the distance component. The normalized peak intensity of the i-th channel is divided by the peak intensity criterion to obtain the intensity component. The distance component and the intensity component are multiplied to obtain the spectral independent resolution of the i-th channel. The missing side distances of the leftmost and rightmost channels are taken as positive infinity. The normalized peak intensity is the ratio of the peak reflectance of this channel to the global maximum reflectance of the current frame, where i is the channel identifier.

5. The method for demodulating and monitoring multi-channel strain signals of valves based on fiber Bragg gratings according to claim 4, characterized in that, The structural coordination deviation is calculated as follows: channels with spectral independent resolution not lower than the resolution threshold are included in the reliable channel set. The measured wavelength of each channel in the reliable channel set is added to the difference between its historical reference wavelength and that of the i-th channel to obtain the single-source prediction value of each reliable channel for the i-th channel. The structural coordination prediction wavelength of the i-th channel is obtained by weighting all single-source prediction values ​​with structural coordination weight. The structural coordination deviation of the i-th channel is obtained by dividing the absolute value of the difference between the spectral reference value of the i-th channel and the structural coordination prediction wavelength by the average full width at half maximum (FWHM). The spectral reference value of the channel in the reliable channel set is the center wavelength of the measured peak obtained by fitting it, and the spectral reference value of the channel in the channel set to be constrained is its spectral indication wavelength.

6. The method for demodulating and monitoring multi-channel strain signals of valves based on fiber Bragg gratings according to claim 5, characterized in that, The rules for classifying each channel into its corresponding diagnostic quadrant are as follows: when the spectral independent resolution is not lower than the resolution threshold and the structural coordination deviation does not exceed the deviation threshold, it is classified into quadrant I; when the spectral independent resolution is not lower than the resolution threshold and the structural coordination deviation exceeds the deviation threshold, it is classified into quadrant II; when the spectral independent resolution is lower than the resolution threshold and the structural coordination deviation does not exceed the deviation threshold, it is classified into quadrant III; and when the spectral independent resolution is lower than the resolution threshold and the structural coordination deviation exceeds the deviation threshold, it is classified into quadrant IV. The two-factor diagnostic composite value is calculated as follows: the quotient obtained by dividing the independent spectral resolution by the resolution threshold is limited to no more than 1, then 1 is subtracted and the square is taken. This square is then added to the square of the quotient obtained by dividing the structural co-deviation by the deviation threshold. The sum of the two is the two-factor diagnostic composite value.

7. The method for demodulating and monitoring multi-channel strain signals of valves based on fiber Bragg gratings according to claim 1, characterized in that, The rule for split processing is: the channel diagnosed as Quadrant I directly uses its measured peak center wavelength as the demodulated output value; The channel diagnosed as Quadrant II will have its measured peak center wavelength used as the demodulated output value and an additional structural anomaly marker; the channels diagnosed as Quadrant III and Quadrant IV will form a subset of constrained solution channels and enter the joint constraint optimization solution.

8. The method for demodulating and monitoring multi-channel strain signals of valves based on fiber Bragg gratings according to claim 7, characterized in that, The overall constraint strength is equal to the product of the constraint strength fundamental coefficient, the spectral factor modulation function value, and the structural factor modulation function value. The spectral factor modulation function is calculated by dividing the spectral independent resolvability by the resolvability determination threshold, limiting the quotient to no more than 1, subtracting 1 from the quotient, and then squaring the result to obtain the spectral factor modulation function value. This function value reflects the impact of the unreliability of spectral data on the constraint requirements. The structural factor modulation function is calculated by dividing the structural coordination deviation by the deviation judgment threshold, taking the square, adding 1 to the squared value, and taking the reciprocal to obtain the structural factor modulation function value. This function value reflects the influence of the reliability of structural coordination prediction on the constraint effectiveness.

9. The method for demodulating and monitoring multi-channel strain signals of valves based on fiber Bragg gratings according to claim 8, characterized in that, The objective function consists of two parts: a data fidelity term and a structural coordination constraint term. The data fidelity term is the sum of squares of the differences between the wavelength to be solved and the spectral indication wavelength of each channel in the constraint solution channel subset. The structural coordination constraint term is the weighted sum of the squares of the wavelength differences between each channel in the constraint solution channel subset and the other channels in all N channels, which deviate from the corresponding historical reference wavelength. The weight is the product of the comprehensive constraint strength of the channel and the structural coordination weight of the corresponding channel pair. Channels belonging to the constraint solution channel subset take the wavelength to be solved, while channels not belonging to the constraint solution channel subset take the determined demodulation output value. By taking the partial derivatives of each wavelength to be solved in the objective function and setting the partial derivatives to zero, a system of linear equations of order K is obtained, where K is the number of channels in the constrained solution channel subset. Solving this system of linear equations yields the cooperatively constrained demodulation wavelength.

10. The method for demodulating and monitoring multi-channel strain signals of valves based on fiber Bragg gratings according to claim 9, characterized in that, The targeted refinement method is as follows: Based on the demodulation wavelength of each channel and the obtained normalized peak intensity and average full width at half maximum of each channel, the theoretical reflection spectrum is reconstructed according to the Gaussian superposition model. The absolute value of the difference between the measured spectrum and the reconstructed spectrum at the demodulation wavelength of each channel is calculated and divided by the global maximum reflection intensity of the current frame to obtain the normalized residual energy. For channels whose normalized residual energy exceeds the residual energy judgment threshold, the refinement step size coefficient is multiplied by the ratio of the normalized residual energy of the channel to the global maximum reflection intensity of the current frame, and then multiplied by the weighted sum of the wavelength difference between the channel and all other channels deviating from the corresponding historical reference wavelength, with the structural synergy weight as the weight, to obtain the correction amount. The correction amount is superimposed on the demodulation wavelength to obtain the refined wavelength. If the normalized residual energy recalculated after refinement is greater than the normalized residual energy before refinement, then the refinement is cancelled and the original demodulation wavelength is restored.

Images