Method and device for evaluating vibration risk level of compressor pipeline

Abstract

The application discloses a kind of vibration risk level evaluation method and device of compressor pipeline, it is related to safety monitoring and intelligent diagnosis technical field.The method comprises: collecting the vibration response data and fluid parameter data of compressor pipeline, and classifying to obtain multi-source vibration dataset;Time delay calculation and phase alignment processing are carried out by reference measuring point sequence, to generate standardized sample;Standardized sample is processed by frequency domain conversion, to obtain multi-source excitation characteristic vector, input multi-source excitation characteristic vector into multi-source data fusion model, to obtain mechanical excitation contribution ratio, fluid excitation contribution ratio and coupling enhancement index;Determine the type of dominant excitation source, retrieve and lock risk frequency band, obtain risk frequency band vibration intensity and compare with standard vibration intensity level, output target vibration risk level.The application uses data-driven mode, without complex physical modeling, realizes multi-source excitation accurate decoupling and quantitative evaluation, improves risk diagnosis accuracy.

Description

Technical Field

[0001] This application relates to the field of safety monitoring and intelligent diagnostic technology, and in particular to a method and apparatus for assessing the vibration risk level of a compressor pipeline. Background Technology

[0002] With the rapid development of industries such as natural gas gathering and transportation, petrochemicals, and coal chemicals, reciprocating compressors and their external pipeline systems have become core and critical equipment in industrial stations, operating in a complex environment of combined mechanical vibration and fluid pulsation. Compressor vibration is transmitted to the pipeline structure through the foundation and connection nodes, creating broadband mechanical excitation. Simultaneously, the periodic intake and exhaust processes induce significant pressure pulsations, flow fluctuations, and turbulent disturbances within the pipeline, creating fluid excitation. The superposition of these excitations can easily lead to vibration amplification in the pipeline system within specific frequency bands, thereby inducing serious engineering safety risks such as support loosening, weld fatigue, and even localized failure, directly threatening the stable operation of the station.

[0003] Currently, the analysis and evaluation of combined vibration problems in compressor piping systems mainly rely on complex physical modeling and simulation (such as finite element analysis) or expensive field test calibration. However, in practical engineering applications, these methods have significant limitations: on the one hand, the generation mechanisms and propagation characteristics of mechanical vibration signals and fluid pulsation signals differ significantly, and existing signal processing methods mostly target single physical quantities, making it difficult to conduct comprehensive analysis under a unified feature space; on the other hand, when mechanical and fluid excitations act together, the system often exhibits nonlinear coupling effects such as resonance amplification, and simple linear superposition models cannot accurately describe the actual physical process, leading to significant deviations in the contribution evaluation results of traditional methods.

[0004] Therefore, how to construct a data-driven intelligent evaluation method that does not rely on complex physical modeling, and achieves accurate decoupling of multi-source incentives, contribution quantification, and efficient risk level determination, is a key problem that urgently needs to be solved in the current field. Summary of the Invention

[0005] In view of this, the vibration risk level assessment method and apparatus for compressor pipelines provided in this application adopts a data-driven mode, eliminating the need for complex physical modeling, achieving accurate decoupling and quantitative assessment of multi-source excitations, improving the accuracy of risk diagnosis, and facilitating its application in industrial settings. The vibration risk level assessment method and apparatus for compressor pipelines provided in this application are implemented as follows: This application provides a method for assessing the vibration risk level of a compressor pipeline, including: Vibration response data and fluid parameter data of compressor pipeline are acquired, and the vibration response data and fluid parameter data are classified to obtain a multi-source vibration dataset, which includes mechanical single-source data, fluid single-source data and dual-source coupled data. A reference measurement point sequence is obtained, and time delay calculation and phase alignment processing are performed on the multi-source vibration dataset and the reference measurement point sequence to obtain standardized samples. The standardized samples are subjected to frequency domain transformation to obtain multi-source excitation feature vectors; A multi-source data fusion model is constructed, and the multi-source excitation feature vector is input into the multi-source data fusion model to obtain the contribution ratio of mechanical excitation, the contribution ratio of fluid excitation, and the coupling enhancement index. The dominant excitation source type is obtained by performing a dominant excitation source determination process on the mechanical excitation contribution ratio, the fluid excitation contribution ratio, and the coupling enhancement index; The risk frequency band is obtained by searching and determining the safety threshold of the dominant excitation source type. Vibration response data of compressor pipelines corresponding to all frequency points within the risk frequency band are extracted. Vibration intensity is obtained based on the vibration response data. The vibration intensity is then compared with the standard vibration intensity level to obtain the target vibration risk level.

[0006] In some embodiments, obtaining a reference measurement point sequence involves performing time delay calculations and phase alignment processing on the multi-source vibration dataset and the reference measurement point sequence to obtain standardized samples, including: The vibration signal of the compressor pipeline is acquired, and the vibration signal is used as a reference measurement point sequence; The time delay estimate is obtained by performing a generalized cross-correlation algorithm and frequency domain weighting on the multi-source vibration dataset and the reference measurement point sequence. The time delay estimate is processed by time axis shifting, anti-aliasing fractional time delay filter construction, and resampling interpolation to obtain the processed time delay estimate. Based on the processed time delay estimate, the multi-source vibration dataset is phase aligned to obtain standardized samples.

[0007] In some embodiments, the step of performing frequency domain transformation on the standardized samples to obtain multi-source excitation feature vectors includes: The standardized samples are subjected to frequency domain transformation to obtain frequency domain data, which includes mechanical single-source frequency domain data, fluid single-source frequency domain data, and dual-source coupled frequency domain data. Distance calculation is performed on the complex spectral vectors of the mechanical single-source frequency domain data, the fluid single-source frequency domain data, and the dual-source coupled frequency domain data to obtain differentiated features; The correlation characteristics are obtained by calculating the average amplitude of the mechanical excitation phase corresponding to the mechanical single-source frequency domain data and the fluid excitation phase corresponding to the fluid single-source frequency domain data. The power spectral density of the dual-source coupled frequency domain data is determined, and the ratio of this power spectral density to the sum of the power spectral densities of the mechanical single-source frequency domain data and the fluid single-source frequency domain data is used to obtain the cooperative characteristics. The differential features, the correlation features, and the collaborative features are integrated to obtain a multi-source excitation feature vector.

[0008] In some embodiments, the process of determining the dominant excitation source type by analyzing the contribution ratio of the mechanical excitation, the contribution ratio of the fluid excitation, and the coupling enhancement index includes: Obtain the mechanical dominance threshold, fluid dominance threshold, and coupling cooperation threshold; By comparing the mechanical excitation contribution ratio with the mechanical dominance threshold, when the mechanical excitation contribution ratio reaches the mechanical dominance threshold, the dominant excitation source type is determined to be the mechanical vibration dominant type. Alternatively, when the mechanical excitation contribution ratio does not reach the mechanical dominance threshold, the fluid excitation contribution ratio is compared with the fluid dominance threshold. When the fluid excitation contribution ratio reaches the fluid dominance threshold, the dominant excitation source type is determined to be the fluid pulsation dominant type. Alternatively, when the contribution ratio of mechanical excitation does not reach the mechanical dominance threshold and the contribution ratio of fluid excitation does not reach the fluid dominance threshold, the coupling enhancement index is compared with the coupling synergy threshold. When the coupling enhancement index reaches the coupling synergy threshold, the dominant excitation source type is determined to be the fluid-structure interaction resonance dominant type.

[0009] In some embodiments, the construction of the multi-source data fusion model, which involves inputting the multi-source excitation feature vectors into the multi-source data fusion model to obtain the mechanical excitation contribution ratio, the fluid excitation contribution ratio, and the coupling enhancement index, includes: An initial multi-source data fusion model is constructed, which is used to establish a nonlinear mapping relationship between the multi-source incentive feature vectors and the true contribution labels; Historical frequency domain data corresponding to mechanical single-source data, fluid single-source data, and dual-source coupled data collected in the past are obtained. Differential features, correlation features, and collaborative features are extracted from the historical frequency domain data. An input feature vector is constructed based on the differential features, correlation features, and collaborative features. A training sample set is constructed based on the input feature vector and the true contribution label; Based on the training sample set, the initial multi-source data fusion model is iteratively optimized to obtain the multi-source data fusion model. Obtain multi-source excitation feature vectors and input them into the multi-source data fusion model to obtain the contribution ratio of mechanical excitation, the contribution ratio of fluid excitation, and the coupling enhancement index.

[0010] In some embodiments, the process of searching and determining the safety threshold for the dominant incentive source type to obtain the risk frequency band includes: Based on the frequency domain data corresponding to the standardized samples, the energy peak distribution range of the dominant excitation source type in the frequency domain is retrieved; A preset safety threshold is determined, and the spectral amplitude value corresponding to each frequency point within the energy peak distribution range is compared with the preset safety threshold to obtain the processed spectral amplitude value. The frequency range exceeding the preset safety threshold in the processed spectral amplitude is locked to obtain the risk frequency band.

[0011] In some embodiments, the mechanical dominance threshold and the fluid dominance threshold are set to 0.60, and the coupling synergy threshold is set to 1.15.

[0012] This application provides a vibration risk level assessment device for compressor pipelines, comprising: The acquisition module is used to acquire vibration response data and fluid parameter data of the compressor pipeline, classify the vibration response data and fluid parameter data to obtain a multi-source vibration dataset, which includes mechanical single-source data, fluid single-source data and dual-source coupled data. The acquisition module is also used to acquire a reference measurement point sequence, perform time delay calculation and phase alignment processing on the multi-source vibration dataset and the reference measurement point sequence to obtain standardized samples; The processing module is used to perform frequency domain transformation on the standardized samples to obtain multi-source excitation feature vectors; A construction module is used to construct a multi-source data fusion model. The multi-source excitation feature vector is input into the multi-source data fusion model to obtain the contribution ratio of mechanical excitation, the contribution ratio of fluid excitation, and the coupling enhancement index. The processing module is further used to perform dominant excitation source determination processing on the mechanical excitation contribution ratio, the fluid excitation contribution ratio, and the coupling enhancement index to obtain the dominant excitation source type; The processing module is also used to perform retrieval and safety threshold determination processing on the dominant excitation source type to obtain the risk frequency band; The processing module is also used to extract vibration response data of compressor pipelines corresponding to all frequency points within the risk frequency band, obtain vibration intensity based on the vibration response data, compare the vibration intensity with the standard vibration intensity level, and obtain the target vibration risk level.

[0013] The computer device provided in this application includes a memory and a processor. The memory stores a computer program that can run on the processor. When the processor executes the program, it implements the method described in this application.

[0014] The computer-readable storage medium provided in this application embodiment stores a computer program thereon, which, when executed by a processor, implements the method described in this application embodiment.

[0015] This application provides a method and apparatus for assessing the vibration risk level of compressor pipelines. It collects vibration response data and fluid parameter data from the compressor pipelines, classifying them into a multi-source vibration dataset. By performing time delay calculations and phase alignment processing using a reference measurement point sequence, standardized samples are generated. The standardized samples undergo frequency domain transformation to obtain multi-source excitation feature vectors. These feature vectors are then input into a multi-source data fusion model to obtain the contribution ratio of mechanical excitation, the contribution ratio of fluid excitation, and the coupling enhancement index. The dominant excitation source type is determined, the risk frequency band is retrieved and locked, the vibration intensity of the risk frequency band is obtained and compared with the standard vibration intensity level, and the target vibration risk level is output. This data-driven approach, without the need for complex physical modeling, achieves accurate decoupling and quantitative assessment of multi-source excitations, improves the accuracy of risk diagnosis, facilitates industrial field application, and solves the technical problems mentioned in the background art. Attached Figure Description

[0016] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments of this application or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 A schematic diagram illustrating the implementation process of a vibration risk level assessment method for compressor pipelines provided in this application embodiment; Figure 2 A schematic diagram illustrating an implementation process for obtaining standardized samples, provided in an embodiment of this application; Figure 3This is a schematic diagram of a vibration risk level assessment device for a compressor pipeline provided in an embodiment of this application. Detailed Implementation

[0018] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0019] The following description of some technologies involved in the embodiments of this application is provided to aid understanding and should be considered merely exemplary. Therefore, those skilled in the art should recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of this application. Similarly, for clarity and brevity, some descriptions of well-known functions and structures are omitted in the following description.

[0020] Figure 1 This is a schematic flowchart illustrating the implementation of a vibration risk level assessment method for compressor pipelines provided in this application, including steps 101 to 107. Wherein, Figure 1 This is merely one execution order shown in the embodiments of this application and does not represent the only execution order for a method of assessing the vibration risk level of compressor pipelines. Where the final result can be achieved, Figure 1 The steps shown can be performed in parallel or in reverse order.

[0021] Step 101: Obtain vibration response data and fluid parameter data of the compressor pipeline, classify the vibration response data and fluid parameter data to obtain a multi-source vibration dataset.

[0022] In this embodiment, firstly, basic data of the compressor pipeline is collected using the station's distributed control system and a portable vibration analyzer. The collected data includes vibration response data from key pipeline nodes (such as the compressor outlet, elbows, and manifolds) and fluid parameter data within the pipeline. Vibration response data is collected using an accelerometer, while fluid parameter data is collected using a pressure sensor or flow meter. During data acquisition, the sampling frequency for both vibration response data and fluid parameter data is set to at least twice the highest frequency of the target analysis frequency band. Furthermore, the sampling frequency range for vibration response data covers the compressor pipeline's fundamental operating frequency and the preset order of higher harmonic frequencies, ensuring data integrity and validity.

[0023] Subsequently, based on the operating load status and airflow pressure status of the compressor pipeline, combined with process operation records, the collected data were classified and processed to construct a multi-source vibration dataset. Data showing the compressor pipeline operating under no-load conditions with no airflow load and only mechanical vibration was transmitted was classified as mechanical single-source data. Data showing airflow passing through the compressor pipeline but the compressor pipeline being stopped or in a non-excited state, with fluid pulsation dominating, was classified as fluid single-source data. Data showing the compressor pipeline operating under normal load, with both mechanical vibration and airflow pulsation acting simultaneously, was classified as dual-source coupled data. These three types of data together constitute the multi-source vibration dataset.

[0024] Step 102: Obtain the reference measurement point sequence, perform time delay calculation and phase alignment processing on the multi-source vibration dataset and the reference measurement point sequence to obtain standardized samples.

[0025] In this embodiment, the vibration signal at the compressor outlet is selected as the reference measurement point sequence. The reference measurement point sequence is highly sensitive to changes in operating conditions and can provide a stable reference benchmark for aligning data from multiple operating conditions.

[0026] For the working condition data in the multi-source vibration dataset, a generalized cross-correlation algorithm based on phase transform weighting is used for processing. The cross-power spectrum between the working condition data and the reference measurement point sequence is calculated, and weighting coefficients are introduced to whiten the cross-power spectrum. The generalized cross-correlation function is obtained through inverse Fourier transform, and its formula is: ,in, For generalized cross-correlation function, For time delay, For the inverse Fourier transform operator, It is a frequency-weighted function. For cross power spectral density, Let x and y be complex exponential basis functions, and x and y be vibration signals.

[0027] By searching for the peak position of the generalized cross-correlation function, the time delay estimate containing both integer and fractional parts is accurately obtained, effectively suppressing the interference of strong background noise and multipath reverberation on site.

[0028] For the integer part, alignment is performed using a time-axis shift method. If the integer part is positive, it indicates that the data sequence to be aligned is lagging; the time index of the integer part is shifted forward by the corresponding number of sampling points, and the header data is discarded. If it is negative, it indicates that the data sequence to be aligned is ahead; the integer part is shifted backward and zeros are added at the beginning. For the fractional part, an anti-aliasing fractional delay filter is constructed, and subsampling delay compensation is achieved through resampling interpolation. The formula is as follows: ,in, The aligned signal is a new signal sequence after phase correction, which serves as the input for subsequent feature extraction; This is a discrete-time index, representing the current calculation of the [number]th [time]. One sampling point; For the convolution summation window, since computers cannot calculate infinity, a calculation range must be set; The original input signal is represented by sampled values ​​of the input signal at different times. For the Sinc function; This is the fractional delay.

[0029] By combining the results of integer time delay alignment and subsampling-level time delay compensation, phase alignment processing of the multi-source vibration dataset in the time dimension is performed to achieve subsampling-level phase locking of multi-condition data, and finally obtain standardized samples.

[0030] Step 103: Perform frequency domain transformation on the standardized samples to obtain multi-source excitation feature vectors.

[0031] In this embodiment of the application, the obtained standardized samples are subjected to frequency domain transformation processing to convert the time series data into frequency domain data. The frequency domain data includes three types: mechanical single-source frequency domain data, fluid single-source frequency domain data, and dual-source coupled frequency domain data, which correspond to the three types of data in the multi-source vibration dataset, respectively.

[0032] Multidimensional features are extracted based on three types of frequency domain data: First, differential features, obtained by calculating the Euclidean distance between the complex spectral vectors of the dual-source coupled frequency domain data and the sum of the complex spectral vectors of the mechanical single-source frequency domain data and the fluid single-source frequency domain data. This is used to characterize the deviation between the actual response under dual-source conditions and the linear superposition response of a single source. Second, correlation features, calculated within a statistical time window, using the average amplitude of the unit complex vector of the difference between the mechanical excitation phase corresponding to the mechanical single-source frequency domain data and the fluid excitation phase corresponding to the fluid single-source frequency domain data, i.e., the spectral phase lock value, to characterize the phase synchronization stability of the two in the time dimension. Third, synergistic features, calculated by the ratio of the power spectral density of the dual-source coupled frequency domain data to the sum of the power spectral densities of the mechanical single-source frequency domain data and the fluid single-source frequency domain data, and then performing a logarithmic transformation on this ratio to obtain the interactive energy gain, used to quantify the energy level amplification factor under coupling. The extracted differential, correlation, and synergistic features are integrated and processed to construct a multi-source excitation feature vector.

[0033] Step 104: Construct a multi-source data fusion model by inputting the multi-source excitation feature vector into the multi-source data fusion model to obtain the contribution ratio of mechanical excitation, the contribution ratio of fluid excitation, and the coupling enhancement index.

[0034] In this embodiment, support vector regression or multilayer perceptron is used to construct an initial multi-source data fusion model. The initial multi-source data fusion model is used to establish a nonlinear mapping relationship between multi-source incentive feature vectors and true contribution labels.

[0035] Historical frequency domain data corresponding to mechanical single-source data, fluid single-source data, and dual-source coupled data are acquired. Differential features, correlation features, and collaborative features are extracted from the historical frequency domain data using feature extraction methods. These three types of features are then vectorized and concatenated to construct the input feature vector required for model training. The formula is as follows: ,in, Represents the fused feature vector; It represents the difference characteristics, reflecting the amplitude difference between two signals in the frequency domain, and its physical meaning corresponds to the energy increment brought about by fluid pulsation; It represents the correlation characteristics, reflecting the degree of linear correlation between two signals in the frequency domain, and its physical meaning is to determine whether the vibrations originate from the same source; It indicates the cooperative / synchronous characteristics, reflecting whether there is a nonlinear "resonance" or "frequency locking" phenomenon in the fluid and mechanical structure; The characteristic is a frequency domain characteristic. It is a vector containing values ​​for each frequency band from low to high frequencies.

[0036] The true contribution under various historical operating conditions is obtained through numerical simulation or controlled experiment calibration. The true contribution is used as the output label and combined with the input feature vector to form a complete training sample set.

[0037] A supervised learning algorithm is used to iteratively optimize the model parameters on the training sample set. During the training process, the goal is to minimize the mean square error between the predicted contribution and the actual contribution. A regularization term is introduced to prevent the model from overfitting until the loss function converges, thus completing the training of the multi-source data fusion model.

[0038] The multi-source excitation feature vectors extracted in real time are input into the trained multi-source data fusion model. The nonlinear fitting capability of the model is used for mapping processing, and the contribution ratio of mechanical excitation, the contribution ratio of fluid excitation, and the coupling enhancement index are obtained analytically within the target frequency band.

[0039] Step 105: Perform dominant excitation source determination processing on the contribution ratio of mechanical excitation, the contribution ratio of fluid excitation, and the coupling enhancement index to obtain the dominant excitation source type.

[0040] In this embodiment, the threshold for determining the dominant excitation source is determined based on the statistical confidence interval of historical normal operation data. Both the mechanical dominance threshold and the fluid dominance threshold are set to 0.60, determined according to the absolute dominance principle where the energy share of a single excitation source exceeds 50% with an additional 10% confidence margin. The coupling and coordination threshold is set to 1.15, determined based on the nonlinear gain criterion that the gain of the dual-source response amplitude relative to the linear superposition amplitude exceeds 15% and that this gain is greater than three standard deviations of the system background noise.

[0041] When the contribution of mechanical excitation is not less than 0.60, the dominant excitation source type is determined to be mechanical vibration dominant type; when the contribution of mechanical excitation is less than 0.60 and the contribution of fluid excitation is not less than 0.60, the dominant excitation source type is determined to be fluid pulsation dominant type; when the contribution of both mechanical excitation and fluid excitation is less than 0.60 and the coupling enhancement index is not less than 1.15, the dominant excitation source type is determined to be fluid-structure interaction resonance dominant type.

[0042] Step 106: Search for the dominant incentive source type and determine the safety threshold to obtain the risk frequency band.

[0043] In this embodiment of the application, based on the frequency domain data corresponding to the standardized samples, the energy peak distribution range of the determined dominant excitation source type in the frequency domain is retrieved, and the frequency range of the dominant excitation source energy concentration is clarified.

[0044] A preset safety threshold is determined, based on statistical analysis of historical normal operation data or industry safety standards. The spectral amplitude corresponding to each frequency point within the energy peak distribution range is compared with the preset safety threshold one by one. Frequency ranges with spectral amplitudes exceeding the preset safety threshold are selected and locked to obtain the risk frequency band.

[0045] Step 107: Extract the vibration response data of the compressor pipeline corresponding to all frequency points within the risk frequency band, obtain the vibration intensity based on the vibration response data, compare the vibration intensity with the standard vibration intensity level, and obtain the target vibration risk level.

[0046] In this embodiment, vibration intensity is calculated for vibration signals within the locked risk frequency band, specifically covering vibration response data corresponding to all frequency points within the band. The calculated vibration intensity is compared with a specified vibration intensity level, and a target vibration risk level is output based on the comparison result. The target vibration risk level includes a warning level, an alarm level, and a vehicle scramble level.

[0047] This application's embodiments, based on multi-source vibration datasets, overcome the limitations of traditional single-data type analysis, effectively decoupling mechanical excitation and fluid pulsation, thus solving the engineering pain point of difficulty in identifying the causes of complex vibrations. The entire process is data-driven, eliminating the need for high-precision pipeline physical structure models or complex simulations, reducing on-site deployment difficulty and computational costs, and facilitating application in industrial sites such as natural gas gathering and transportation and petrochemical plants. Integrating key processes such as phase alignment, feature extraction, intelligent model solving, and risk assessment, it can output multi-dimensional evaluation results including the dominant excitation source, risk frequency band, and vibration risk level, providing comprehensive and high-precision data support for fault warning and operation and maintenance decisions.

[0048] In the above Figure 1 Based on the above, this application embodiment also provides a schematic diagram of the implementation process for obtaining standardized samples. For example... Figure 2 As shown, steps 201 to 204 are included: Step 201: Obtain the vibration signal of the compressor pipeline and use the vibration signal as a reference measurement point sequence.

[0049] In this embodiment, a reference measurement point sequence is obtained. Based on the operating characteristics of the compressor pipeline, the vibration signal at the compressor outlet, which is highly sensitive to changes in operating conditions and exhibits strong signal stability, is selected as the reference measurement point sequence. This measurement point can provide a reliable benchmark for phase alignment of multi-condition data, ensuring the consistency and accuracy of subsequent registration operations.

[0050] Step 202: Perform generalized cross-correlation algorithm processing and frequency domain weighting processing on the multi-source vibration dataset and the reference measurement point sequence to obtain the time delay estimate.

[0051] In this embodiment, the mechanical single-source data, fluid single-source data, and dual-source coupled data in the multi-source vibration dataset are processed separately with a reference measurement point sequence. A generalized cross-correlation algorithm based on phase transform weighting is used to calculate the cross-power spectrum between the various working condition data sequences and the reference measurement point sequence. At the same time, frequency domain weighting coefficients are introduced to whiten the cross-power spectrum, effectively suppressing strong background noise in industrial sites and multipath reverberation interference caused by pipeline structures, sharpening the peak value of the cross-correlation function, thereby accurately obtaining the time delay estimate containing both integer and fractional parts, avoiding the problem of insufficient alignment accuracy caused by poor correlation peak characteristics in traditional cross-correlation algorithms.

[0052] Step 203 involves performing time axis shifting, constructing an anti-aliasing fractional delay filter, and resampling interpolation on the time delay estimate to obtain the processed time delay estimate.

[0053] In this embodiment, the integer part of the time delay estimate is processed using a time axis shifting method. If the integer part is positive, it indicates that the data sequence to be aligned lags behind the reference measurement point sequence. The time index of the data sequence is shifted forward by the corresponding number of sampling points, and redundant data at the beginning is discarded. If the integer part is negative, it indicates that the data sequence to be aligned is ahead of the reference measurement point sequence. The data sequence is shifted backward, and zeros are padded at the beginning to ensure consistent data length. For the fractional part of the time delay estimate, an anti-aliasing fractional time delay filter is constructed. Subsampling-level time delay compensation is achieved through resampling interpolation, solving the problem that traditional shifting methods cannot handle fractional time delays and improving the accuracy of time delay processing.

[0054] Step 204: Based on the processed time delay estimates, perform phase alignment processing on the multi-source vibration dataset to obtain standardized samples.

[0055] In this embodiment, phase alignment is performed and standardized samples are generated. The time axis translation result of the integer part is combined with the subsampling-level time delay compensation result of the fractional part to perform phase alignment processing on the multi-source vibration dataset in the time dimension, realizing subsampling-level phase locking of multi-condition data, ensuring that the mechanical vibration waveform and the fluid pulsation waveform are strictly synchronized in time scale, and finally obtaining the time-registered standardized samples.

[0056] This application's embodiments achieve subsampling-level phase locking by processing the integer part of the time delay through time axis shifting and processing the fractional part through an anti-aliasing fractional time delay filter. This ensures strict synchronization of multi-condition data on the time scale, providing a consistent and high-quality data foundation for subsequent frequency domain feature extraction and data fusion. The phase alignment process is standardized and highly operable, requiring no manual intervention or adjustment, thus improving the automation level and stability of the evaluation method and avoiding problems such as distortion in subsequent feature extraction and deviation in model solution caused by data asynchrony.

[0057] In some embodiments, frequency domain transformation processing is performed on standardized samples to obtain multi-source excitation feature vectors, including: frequency domain transformation processing is performed on standardized samples to obtain frequency domain data, the frequency domain data including mechanical single-source frequency domain data, fluid single-source frequency domain data and dual-source coupled frequency domain data.

[0058] Specifically, the standardized samples are transformed into frequency domain data. A frequency domain transformation algorithm is used to convert the phase-aligned standardized samples from time-series data into frequency domain data. The frequency domain data corresponds to the classification of the multi-source vibration dataset, specifically including mechanical single-source frequency domain data, fluid single-source frequency domain data, and dual-source coupled frequency domain data, which correspond to the frequency domain representation of mechanical single-source data, fluid single-source data, and dual-source coupled data, respectively.

[0059] Furthermore, distance calculations are performed on the complex spectral vectors of mechanical single-source frequency domain data, fluid single-source frequency domain data, and dual-source coupled frequency domain data to obtain differentiated features.

[0060] Specifically, in the complex frequency domain, Euclidean distance is calculated for the sum of the complex spectral vectors of the dual-source coupled frequency domain data and the complex spectral vectors of the mechanical single-source frequency domain data and the fluid single-source frequency domain data. The Euclidean distance, also known as the nonlinear vector residual, is the differential characteristic used to accurately characterize the deviation between the actual response under dual-source conditions and the linear superposition of the mechanical and fluid single-source responses, reflecting the nonlinear coupling effect of the system.

[0061] Furthermore, the average amplitude of the mechanical excitation phase corresponding to the mechanical single-source frequency domain data and the fluid excitation phase corresponding to the fluid single-source frequency domain data is calculated to obtain the correlation characteristics.

[0062] Specifically, a statistical time window is set, and within this window, the mechanical excitation phase corresponding to the single-source mechanical frequency domain data and the fluid excitation phase corresponding to the single-source fluid frequency domain data are extracted. The unit complex vector of the phase difference between the two is calculated, and then the average amplitude of the unit complex vector is calculated. The result is the correlation feature, also known as the spectral phase lock value, which is used to characterize the phase synchronization stability of mechanical and fluid excitations in the time dimension. The higher the degree of phase synchronization, the greater the risk of resonance.

[0063] Furthermore, the power spectral density of the dual-source coupled frequency domain data is determined, and the ratio of this ratio to the sum of the power spectral densities of the mechanical single-source frequency domain data and the fluid single-source frequency domain data is used to obtain the cooperative characteristics.

[0064] Specifically, the power spectral density of the single-source mechanical data and the single-source fluid data are extracted separately, and their sum is calculated. Simultaneously, the power spectral density of the coupled dual-source data is extracted, and the ratio of the coupled dual-source power spectral density to the sum of the two single-source power spectral densities is calculated. This ratio is then logarithmically transformed. The result is the cooperative feature, also known as the interactive energy gain, used to quantify the energy level amplification factor under dual-source coupling.

[0065] Furthermore, the differential features, correlation features, and collaborative features are integrated to obtain a multi-source excitation feature vector.

[0066] Specifically, the obtained differential features, correlation features, and collaborative features are vectorized and concatenated to form a complete multi-source excitation feature vector. The multi-source excitation feature vector can comprehensively reflect the nonlinear relationship, spectral correlation, and coupling gain characteristics of multi-source excitation, providing core feature support for the input of subsequent multi-source data fusion models.

[0067] This application's embodiments transform time-series data into quantifiable frequency-domain features through frequency domain conversion, constructing a multi-dimensional feature space that includes differential, correlated, and synergistic features. This comprehensively captures the nonlinear coupling relationship between multi-source excitations, overcoming the limitation of traditional single features in reflecting complex vibration mechanisms. Differential features accurately quantify the deviation between dual-source responses and single-source linear superposition, correlated features characterize phase synchronization stability, and synergistic features reflect energy coupling gain. These three types of features complement each other, providing comprehensive and effective input for multi-source data fusion models and improving the accuracy of subsequent contribution calculations. The feature extraction logic is clear, and the calculation rules are well-defined. Quantitative analysis based on frequency-domain data avoids interference from subjective experience judgments, ensuring the objectivity and comparability of features and providing the possibility for universal evaluation of different working conditions and equipment.

[0068] In some embodiments, the dominant excitation source is determined by processing the contribution ratio of mechanical excitation, the contribution ratio of fluid excitation, and the coupling enhancement index to obtain the dominant excitation source type, including: obtaining the mechanical dominant threshold, the fluid dominant threshold, and the coupling coordination threshold.

[0069] Specifically, statistical confidence interval analysis was performed based on historical normal operation data of the compressor pipeline to identify three types of judgment thresholds. The mechanical dominance threshold and the fluid dominance threshold were both set to 0.60. This value is determined based on the absolute dominance principle, meaning that after the energy share of a single excitation source exceeds 50%, an additional 10% confidence margin is added to ensure the reliability of the judgment. The coupling coordination threshold was set to 1.15. This value is determined based on the nonlinear gain criterion, requiring that the gain of the dual-source response amplitude relative to the linear superposition amplitude of the single source exceeds 15%, and that the gain is greater than three standard deviations of the system background noise, in order to accurately identify nonlinear coupling effects.

[0070] Furthermore, by comparing the contribution ratio of mechanical excitation with the mechanical dominance threshold, when the contribution ratio of mechanical excitation reaches the mechanical dominance threshold, the dominant excitation source type is determined to be the mechanical vibration dominant type.

[0071] Specifically, the mechanical excitation contribution ratio output by the multi-source data fusion model is extracted and quantitatively compared with a preset mechanical dominance threshold to determine whether the mechanical excitation contribution ratio reaches or exceeds the threshold. If the mechanical excitation contribution ratio is greater than or equal to 0.60, the dominant excitation source type is directly determined to be the mechanical vibration dominant type, indicating that the vibration of the compressor pipeline is mainly caused by the mechanical excitation transmitted by the compressor body.

[0072] Furthermore, when the contribution of mechanical excitation does not reach the mechanical dominance threshold, the contribution of fluid excitation is compared with the fluid dominance threshold. When the contribution of fluid excitation reaches the fluid dominance threshold, the dominant excitation source type is determined to be fluid pulsation dominance type.

[0073] Specifically, when the contribution of mechanical excitation is below the threshold, the contribution of fluid excitation is compared with the fluid dominance threshold. If the comparison shows that the contribution of mechanical excitation is less than 0.60, the contribution of fluid excitation is further extracted and compared with the fluid dominance threshold. If the contribution of fluid excitation is greater than or equal to 0.60, the dominant excitation source type is determined to be fluid pulsation dominant type, indicating that the pipeline vibration is mainly caused by fluid excitation such as pressure pulsation and flow fluctuation in the pipe.

[0074] Furthermore, when the contribution of mechanical excitation does not reach the mechanical dominance threshold and the contribution of fluid excitation does not reach the fluid dominance threshold, the coupling enhancement index and the coupling synergy threshold are compared. When the coupling enhancement index reaches the coupling synergy threshold, the dominant excitation source type is determined to be the fluid-structure interaction resonance dominant type.

[0075] Specifically, when the contribution proportions of both single-source types do not reach the threshold, the coupling enhancement index is compared with the coupling synergy threshold. If the contribution proportion of mechanical excitation is less than 0.60 and the contribution proportion of fluid excitation is less than 0.60, the coupling enhancement index output by the model is extracted and compared with the coupling synergy threshold. If the coupling enhancement index is greater than or equal to 1.15, the dominant excitation source type is determined to be the fluid-structure interaction resonance dominant type, indicating that the pipe vibration is caused by the nonlinear coupling effect generated by the interaction of mechanical excitation and fluid excitation.

[0076] This application's embodiments establish a dominant excitation source determination logic based on quantified thresholds. Through hierarchical comparison of mechanical dominant thresholds, fluid dominant thresholds, and coupling synergy thresholds, it achieves accurate identification of three types of vibration causes: mechanical vibration-dominant, fluid pulsation-dominant, and fluid-structure interaction resonance-dominant, solving the problem of existing technologies' difficulty in clearly identifying the core cause of vibration. The determination process has a rigorous logical progression, first determining the dominant source of vibration and then determining the coupling-dominant source, avoiding ambiguity and ensuring the uniqueness and reliability of the results. The determination results directly point to the core cause of vibration, providing a clear basis for targeted operation and maintenance, reducing the cost waste caused by blind operation and maintenance, and improving the efficiency of equipment fault management.

[0077] In some embodiments, a multi-source data fusion model is constructed, and multi-source excitation feature vectors are input into the multi-source data fusion model to obtain the mechanical excitation contribution ratio, the fluid excitation contribution ratio, and the coupling enhancement index. This includes: constructing an initial multi-source data fusion model, which is used to establish a nonlinear mapping relationship between the multi-source excitation feature vectors and the true contribution labels.

[0078] Specifically, an initial multi-source data fusion model is constructed. Support vector regression or multilayer perceptron is selected as the basic model framework to construct the initial multi-source data fusion model. The core function of the initial multi-source data fusion model is to establish a nonlinear mapping relationship between multi-source activation feature vectors and true contribution labels. The support vector regression model can use radial basis functions as kernel functions to adapt to the nonlinear fitting requirements under complex working conditions.

[0079] Furthermore, historical frequency domain data corresponding to historically collected mechanical single-source data, fluid single-source data, and dual-source coupled data are obtained, and differential features, correlation features, and collaborative features are extracted from the historical frequency domain data. Input feature vectors are constructed based on the differential features, correlation features, and collaborative features.

[0080] Specifically, the input feature vector is constructed. Historically collected single-source mechanical data, single-source fluid data, and dual-source coupled data of compressor pipelines are acquired. Using frequency domain transformation and feature extraction methods, these historical data are converted into corresponding historical frequency domain data. Then, differential features (nonlinear vector residuals), correlation features (spectral phase lock values), and cooperative features (interactive energy gain) are extracted from the historical frequency domain data. The three types of features are vectorized and concatenated to construct the input feature vector required for model training.

[0081] Furthermore, a training sample set is constructed based on the input feature vector and the true contribution label.

[0082] Specifically, a training sample set is constructed. A digital twin model of the compressor pipeline is established using numerical simulation software, or controlled experiments are conducted. Given the known mechanical and fluid excitation parameters, the true contribution of mechanical and fluid excitation to pipeline vibration under various historical operating conditions is calibrated, and this true contribution is used as the output label. The constructed input feature vectors are matched one-to-one with the corresponding output labels to form a complete training sample set. If necessary, the sample set can be preprocessed by normalization to improve the stability of model training.

[0083] Furthermore, the initial multi-source data fusion model is iteratively optimized based on the training sample set to obtain the multi-source data fusion model.

[0084] Specifically, the model parameters are optimized to obtain the final multi-source data fusion model. A supervised learning algorithm is used to train the initial multi-source data fusion model. During training, the goal is to minimize the mean squared error between the predicted contribution and the actual contribution, while a regularization term is introduced to suppress overfitting. The model parameters are iteratively adjusted until the loss function converges. At this point, the model's fitting accuracy and generalization ability meet the preset requirements, completing the training and obtaining the final multi-source data fusion model.

[0085] Furthermore, multi-source excitation feature vectors are obtained and input into the multi-source data fusion model to obtain the contribution ratio of mechanical excitation, the contribution ratio of fluid excitation, and the coupling enhancement index.

[0086] Specifically, the process involves inputting a multi-source excitation feature vector and calculating the target index. Based on real-time collected compressor pipeline vibration response data and fluid parameter data, a real-time multi-source excitation feature vector is obtained after constructing a multi-source vibration dataset, phase alignment, frequency domain transformation, and feature extraction. This vector is then input into a trained multi-source data fusion model. The model's nonlinear fitting capability is used for mapping calculations, and the contribution ratios of mechanical excitation, fluid excitation, and coupling enhancement indices are analytically output within the target frequency band.

[0087] This application's embodiments overcome the limitations of traditional linear superposition models, which cannot fit the coupling laws of multi-source excitation under complex working conditions, by constructing a nonlinear mapping relationship using a multi-source data fusion model. This significantly improves the calculation accuracy of the contribution ratio of mechanical excitation, the contribution ratio of fluid excitation, and the coupling enhancement index. A training sample set is constructed based on historical multi-working-condition data, combined with true contribution labels calibrated by numerical simulation or controlled experiments, ensuring the effectiveness and generalization ability of the model training. This allows the model to adapt to changes in different operating conditions and equipment states. During model training, parameters are optimized through supervised learning algorithms until the loss function converges, demonstrating stable nonlinear fitting capabilities. In the online evaluation stage, only real-time feature vectors need to be input for rapid result output, meeting the needs of online monitoring and real-time evaluation in industrial settings.

[0088] In some embodiments, the dominant excitation source type is retrieved and a safety threshold is determined to obtain a risk frequency band, including: retrieving the energy peak distribution range of the dominant excitation source in the frequency domain based on the frequency domain data corresponding to the standardized samples.

[0089] Specifically, the peak energy distribution range in the frequency domain corresponding to the dominant excitation source type is retrieved. Based on the standardized samples obtained after phase alignment processing, the corresponding frequency domain data is obtained through frequency domain transformation; according to the determined dominant excitation source type, the corresponding target frequency domain data is matched. If it is a mechanical vibration dominant type, the mechanical single-source frequency domain data is used as the retrieval object; if it is a fluid pulsation dominant type, the fluid single-source frequency domain data is used as the retrieval object; if it is a fluid-structure interaction resonance dominant type, the dual-source coupling frequency domain data is used as the retrieval object; through frequency domain energy analysis, the peak distribution range of energy concentration in the target frequency domain data is located, and the frequency range mainly covered by the dominant excitation source energy is determined.

[0090] Furthermore, a preset safety threshold is determined, and the spectral amplitude value corresponding to each frequency point within the energy peak distribution range is compared with the preset safety threshold to obtain the processed spectral amplitude value.

[0091] Specifically, a preset safety threshold is determined and spectral amplitude comparisons are performed. Statistical analysis is conducted based on historical normal operation data of the compressor pipeline to determine the preset safety threshold, which is used to define the boundary between normal energy levels and risky energy levels in the frequency domain. For the retrieved energy peak distribution range, the spectral amplitude corresponding to each frequency point within the range is extracted, and the spectral amplitude of each frequency point is quantitatively compared with the preset safety threshold one by one, recording the relationship between the spectral amplitude of each frequency point and the threshold.

[0092] Furthermore, the frequency range exceeding the preset safety threshold in the processed spectral amplitude is locked to obtain the risk frequency band.

[0093] Specifically, the results are compared and filtered to identify all frequency points whose spectral amplitude exceeds a preset safety threshold. If the frequency points exceeding the threshold are continuously distributed, the continuous frequency range is directly identified as the risk band. If the frequency points exceeding the threshold are discretely distributed, adjacent discrete frequency points are integrated to form a complete risk frequency range and locked in, thus obtaining a clear risk band, providing an accurate analysis range for subsequent vibration intensity calculation and risk level assessment.

[0094] This application's embodiments achieve focused risk analysis by retrieving the energy peak distribution range based on the dominant excitation source, avoiding indiscriminate analysis across the entire frequency range and improving assessment efficiency. The preset safety threshold is set based on historical normal operation data or industry standards, ensuring its rationality and applicability. By comparing the spectral amplitude with the threshold point by point, the frequency range exceeding the safety threshold is accurately identified, solving the problem of ambiguous risk frequency band positioning in existing technologies. The precise identification of the risk frequency band allows subsequent vibration intensity calculations to focus only on high-risk intervals, reducing interference from invalid data and improving the relevance and accuracy of risk level assessment. This provides precise frequency dimension basis for equipment overload warnings and fault prediction, helping to avoid structural damage caused by vibration amplification in advance.

[0095] While this application provides the method operation steps as described in the embodiments or flowcharts, more or fewer operation steps may be included based on conventional or non-inventive labor. The order of steps listed in this embodiment is merely one possible execution order among many and does not represent the only execution order. In actual device or client product execution, the methods shown in this embodiment or the accompanying drawings can be executed sequentially or in parallel (e.g., in a parallel processor or multi-threaded processing environment).

[0096] like Figure 3 As shown in the illustration, this application also provides a vibration risk level assessment device 300 for compressor pipelines. The device includes: The acquisition module 301 is used to acquire vibration response data and fluid parameter data of the compressor pipeline, classify the vibration response data and fluid parameter data to obtain a multi-source vibration dataset, which includes mechanical single-source data, fluid single-source data and dual-source coupled data.

[0097] The acquisition module 301 is also used to acquire a reference measurement point sequence, perform time delay calculation and phase alignment processing on the multi-source vibration dataset and the reference measurement point sequence, and obtain standardized samples.

[0098] The processing module 302 is used to perform frequency domain transformation on the standardized samples to obtain multi-source excitation feature vectors.

[0099] Module 303 is used to construct a multi-source data fusion model. The multi-source excitation feature vector is input into the multi-source data fusion model to obtain the contribution ratio of mechanical excitation, the contribution ratio of fluid excitation, and the coupling enhancement index.

[0100] The processing module 302 is also used to determine the dominant excitation source by processing the contribution ratio of mechanical excitation, the contribution ratio of fluid excitation, and the coupling enhancement index, so as to obtain the dominant excitation source type.

[0101] The processing module 302 is also used to retrieve the dominant excitation source type and determine the safety threshold to obtain the risk frequency band.

[0102] The processing module 302 is also used to extract the vibration response data of the compressor pipeline corresponding to all frequency points within the risk frequency band, obtain the vibration intensity based on the vibration response data, compare the vibration intensity with the standard vibration intensity level, and obtain the target vibration risk level.

[0103] Some modules in the apparatus described in this application can be described in the general context of computer-executable instructions that are executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, classes, etc., that perform a specific task or implement a specific abstract data type. This application can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0104] The apparatus or module described in the above embodiments can be implemented by a computer chip or physical entity, or by a product with a certain function. For ease of description, the above apparatus is described by dividing it into various modules according to their functions. When implementing the embodiments of this application, the functions of each module can be implemented in one or more software and / or hardware. Of course, a module that implements a certain function can also be implemented by combining multiple sub-modules or sub-units.

[0105] The methods, apparatus, or modules described in this application can be implemented in a computer-readable program code manner. The controller can be implemented in any suitable manner, such as a microprocessor or processor and a computer-readable medium storing computer-readable program code (e.g., software or firmware) executable by the (micro)processor, logic gates, switches, application-specific integrated circuits (ASICs), programmable logic controllers, and embedded microcontrollers. Examples of controllers include, but are not limited to, the following microcontrollers: ARC 625D, Atmel AT91SAM, Microchip PIC18F26K20, and Silicon Labs C8051F320. A memory controller can also be implemented as part of the control logic of a memory. Those skilled in the art will also recognize that, in addition to implementing the controller in purely computer-readable program code manner, the same functionality can be achieved by logically programming the method steps to make the controller take the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, and embedded microcontrollers. Therefore, such a controller can be considered a hardware component, and the means included within it for implementing various functions can also be considered as structures within the hardware component. Alternatively, the device used to implement various functions can be viewed as either a software module that implements the method or a structure within a hardware component.

[0106] This application also provides an apparatus, the apparatus comprising: a processor; a memory for storing processor-executable instructions; wherein, when the processor executes the executable instructions, it implements the method described in this application.

[0107] This application also provides a non-volatile computer-readable storage medium storing a computer program or instructions thereon, which, when executed, enables the method described in this application embodiment to be implemented.

[0108] Furthermore, in the various embodiments of the present invention, each functional module can be integrated into a processing module, or each module can exist independently, or two or more modules can be integrated into a single module.

[0109] The aforementioned storage media include, but are not limited to, Random Access Memory (RAM), Read-Only Memory (ROM), Cache, Hard Disk Drive (HDD), or Memory Card. The memory can be used to store computer program instructions.

[0110] As can be seen from the above description of the embodiments, those skilled in the art can clearly understand that this application can be implemented by means of software plus necessary hardware. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product, or it can be embodied in the process of data migration. The computer software product can be stored in a storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, mobile terminal, server, or network device, etc.) to execute the methods described in various embodiments or some parts of the embodiments of this application.

[0111] The various embodiments described in this specification are presented in a progressive manner. Similar or identical parts between embodiments can be referred to interchangeably. Each embodiment focuses on its differences from other embodiments. All or part of this application can be used in numerous general-purpose or special-purpose computer system environments or configurations. Examples include: personal computers, server computers, handheld or portable devices, tablet devices, mobile communication terminals, multiprocessor systems, microprocessor-based systems, programmable electronic devices, network PCs, minicomputers, mainframe computers, and distributed computing environments including any of the above systems or devices, etc.

[0112] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit this application. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of this application.

Claims

1. A method for assessing the vibration risk level of compressor pipelines, characterized in that, include: Vibration response data and fluid parameter data of compressor pipeline are acquired, and the vibration response data and fluid parameter data are classified to obtain a multi-source vibration dataset, which includes mechanical single-source data, fluid single-source data and dual-source coupled data. A reference measurement point sequence is obtained, and time delay calculation and phase alignment processing are performed on the multi-source vibration dataset and the reference measurement point sequence to obtain standardized samples. The standardized samples are subjected to frequency domain transformation to obtain multi-source excitation feature vectors; A multi-source data fusion model is constructed, and the multi-source excitation feature vector is input into the multi-source data fusion model to obtain the contribution ratio of mechanical excitation, the contribution ratio of fluid excitation, and the coupling enhancement index. The dominant excitation source type is obtained by performing a dominant excitation source determination process on the mechanical excitation contribution ratio, the fluid excitation contribution ratio, and the coupling enhancement index; The risk frequency band is obtained by searching and determining the safety threshold of the dominant excitation source type. Vibration response data of compressor pipelines corresponding to all frequency points within the risk frequency band are extracted. Vibration intensity is obtained based on the vibration response data. The vibration intensity is then compared with the standard vibration intensity level to obtain the target vibration risk level.

2. The method according to claim 1, characterized in that, The process of obtaining a reference measurement point sequence involves calculating the time delay and performing phase alignment processing on the multi-source vibration dataset and the reference measurement point sequence to obtain standardized samples, including: The vibration signal of the compressor pipeline is acquired, and the vibration signal is used as a reference measurement point sequence; The time delay estimate is obtained by performing a generalized cross-correlation algorithm and frequency domain weighting on the multi-source vibration dataset and the reference measurement point sequence. The time delay estimate is processed by time axis shifting, anti-aliasing fractional time delay filter construction, and resampling interpolation to obtain the processed time delay estimate. Based on the processed time delay estimate, the multi-source vibration dataset is phase aligned to obtain standardized samples.

3. The method according to claim 1, characterized in that, The step of performing frequency domain transformation on the standardized samples to obtain multi-source excitation feature vectors includes: The standardized samples are subjected to frequency domain transformation to obtain frequency domain data, which includes mechanical single-source frequency domain data, fluid single-source frequency domain data, and dual-source coupled frequency domain data. Distance calculation is performed on the complex spectral vectors of the mechanical single-source frequency domain data, the fluid single-source frequency domain data, and the dual-source coupled frequency domain data to obtain differentiated features; The correlation characteristics are obtained by calculating the average amplitude of the mechanical excitation phase corresponding to the mechanical single-source frequency domain data and the fluid excitation phase corresponding to the fluid single-source frequency domain data. The power spectral density of the dual-source coupled frequency domain data is determined, and the ratio of this power spectral density to the sum of the power spectral densities of the mechanical single-source frequency domain data and the fluid single-source frequency domain data is used to obtain the cooperative characteristics. The differential features, the correlation features, and the collaborative features are integrated to obtain a multi-source excitation feature vector.

4. The method according to claim 1, characterized in that, The process of determining the dominant excitation source by analyzing the contribution ratio of mechanical excitation, the contribution ratio of fluid excitation, and the coupling enhancement index yields the dominant excitation source type, including: Obtain the mechanical dominance threshold, fluid dominance threshold, and coupling cooperation threshold; By comparing the mechanical excitation contribution ratio with the mechanical dominance threshold, when the mechanical excitation contribution ratio reaches the mechanical dominance threshold, the dominant excitation source type is determined to be the mechanical vibration dominant type. Alternatively, when the mechanical excitation contribution ratio does not reach the mechanical dominance threshold, the fluid excitation contribution ratio is compared with the fluid dominance threshold. When the fluid excitation contribution ratio reaches the fluid dominance threshold, the dominant excitation source type is determined to be the fluid pulsation dominant type. Alternatively, when the contribution ratio of mechanical excitation does not reach the mechanical dominance threshold and the contribution ratio of fluid excitation does not reach the fluid dominance threshold, the coupling enhancement index is compared with the coupling synergy threshold. When the coupling enhancement index reaches the coupling synergy threshold, the dominant excitation source type is determined to be the fluid-structure interaction resonance dominant type.

5. The method according to claim 1, characterized in that, The construction of the multi-source data fusion model involves inputting the multi-source excitation feature vectors into the model to obtain the contribution ratio of mechanical excitation, the contribution ratio of fluid excitation, and the coupling enhancement index, including: An initial multi-source data fusion model is constructed, which is used to establish a nonlinear mapping relationship between the multi-source incentive feature vectors and the true contribution labels; Historical frequency domain data corresponding to mechanical single-source data, fluid single-source data, and dual-source coupled data collected in the past are obtained. Differential features, correlation features, and collaborative features are extracted from the historical frequency domain data. An input feature vector is constructed based on the differential features, correlation features, and collaborative features. A training sample set is constructed based on the input feature vector and the true contribution label; Based on the training sample set, the initial multi-source data fusion model is iteratively optimized to obtain the multi-source data fusion model. Obtain multi-source excitation feature vectors and input them into the multi-source data fusion model to obtain the contribution ratio of mechanical excitation, the contribution ratio of fluid excitation, and the coupling enhancement index.

6. The method according to claim 1, characterized in that, The process of searching and determining the safety threshold for the dominant incentive source type to obtain the risk frequency band includes: Based on the frequency domain data corresponding to the standardized samples, the energy peak distribution range of the dominant excitation source type in the frequency domain is retrieved; A preset safety threshold is determined, and the spectral amplitude value corresponding to each frequency point within the energy peak distribution range is compared with the preset safety threshold to obtain the processed spectral amplitude value. The frequency range exceeding the preset safety threshold in the processed spectral amplitude is locked to obtain the risk frequency band.

7. The method according to claim 4, characterized in that, The mechanical dominance threshold and the fluid dominance threshold are set to 0.60, and the coupling cooperation threshold is set to 1.

15.

8. A vibration risk level assessment device for compressor pipelines, characterized in that, include: The acquisition module is used to acquire vibration response data and fluid parameter data of the compressor pipeline, classify the vibration response data and fluid parameter data to obtain a multi-source vibration dataset, which includes mechanical single-source data, fluid single-source data and dual-source coupled data. The acquisition module is also used to acquire a reference measurement point sequence, perform time delay calculation and phase alignment processing on the multi-source vibration dataset and the reference measurement point sequence to obtain standardized samples; The processing module is used to perform frequency domain transformation on the standardized samples to obtain multi-source excitation feature vectors; A construction module is used to construct a multi-source data fusion model. The multi-source excitation feature vector is input into the multi-source data fusion model to obtain the contribution ratio of mechanical excitation, the contribution ratio of fluid excitation, and the coupling enhancement index. The processing module is further used to perform dominant excitation source determination processing on the mechanical excitation contribution ratio, the fluid excitation contribution ratio, and the coupling enhancement index to obtain the dominant excitation source type; The processing module is also used to perform retrieval and safety threshold determination processing on the dominant excitation source type to obtain the risk frequency band; The processing module is also used to extract vibration response data of compressor pipelines corresponding to all frequency points within the risk frequency band, obtain vibration intensity based on the vibration response data, compare the vibration intensity with the standard vibration intensity level, and obtain the target vibration risk level.

9. A computer device comprising a memory and a processor, the memory storing a computer program executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the method according to any one of claims 1 to 7.

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

Images