A life prediction method and system for a noise reduction device used in a substation

Abstract

The present invention relates to the technical field of noise reduction device life prediction, and discloses a life prediction method and system for noise reduction devices used in substations. The method comprises collecting operational data of noise reduction devices within the substation, performing feature extraction, and obtaining aging characteristic parameters; performing cluster analysis on the aging characteristic parameters to obtain sound absorption degradation; constructing a finite element model based on the aging characteristic parameters, performing structural sound transmission simulation, and obtaining simulated sound transmission loss; and obtaining sound transmission gain based on the simulated sound transmission loss and a reference sound transmission loss; inputting the aging characteristic parameters, sound absorption degradation, and sound transmission gain into a device aging degree quantification model to obtain the device aging degree; and calculating the remaining life of the noise reduction device based on the device aging degree using a Weibull distribution. The present invention achieves precise mapping from multi-source signals to aging states, and through accurate aging assessment and life prediction, provides reliable technical support for maintenance decisions for noise reduction devices.

Description

Technical Field

[0001] The present invention relates to the technical field of life prediction of noise reduction devices, and in particular to a life prediction method and system for a noise reduction device used in a transformer substation. Background Art

[0002] Noise reduction devices offer highly effective sound absorption and insulation, effectively reducing noise from substation equipment. The long-term and stable operation of these devices plays a crucial role in the sustainable development of power systems. Accurately assessing the aging of key components is crucial for ensuring safe equipment operation and optimizing maintenance strategies throughout the lifecycle of noise reduction devices.

[0003] Traditional device life prediction relies heavily on single sensor data or empirical strategies based on "scheduled maintenance." The limitations of existing methods are that, on the one hand, single sensor data often indirectly infers overall performance through certain local physical quantities, ignoring the coupling relationship between structure and acoustic properties, resulting in insufficient prediction accuracy. On the other hand, reliance on "fixed-cycle replacement" or "post-failure repair" models can lead to material waste due to excessive maintenance and sudden performance degradation due to delayed maintenance. This extensive maintenance strategy results in insufficient efficiency in controlling costs throughout the entire life cycle. During long-term use, the degradation of materials and structures in noise reduction devices presents a chain reaction of single-factor triggering and multi-factor amplification, further exacerbating the complexity of life prediction. Current prediction methods cannot effectively address the risk of systemic failure caused by multi-physical field coupling during long-term service. Summary of the Invention

[0004] In order to solve the above technical problems, the present invention provides a life prediction method and system for a noise reduction device for a substation, which can solve the problem of insufficient accuracy of existing prediction methods, realize accurate and efficient device aging assessment and life prediction, and provide reliable technical support for maintenance decisions of the noise reduction device.

[0005] In a first aspect, the present invention provides a life prediction method for a noise reduction device for a substation, the method comprising:

[0006] Collecting operating data of a noise reduction device in a substation, performing feature extraction on the operating data, and obtaining aging feature parameters;

[0007] Performing cluster analysis on the aging characteristic parameters, and obtaining the degradation stage and the corresponding sound absorption decay amount according to the clustering results;

[0008] constructing a finite element model of the noise reduction device according to the aging characteristic parameters, performing a structure-borne sound transmission simulation based on the finite element model to obtain a simulated sound transmission loss, and obtaining a sound transmission gain according to the simulated sound transmission loss and a preset reference sound transmission loss;

[0009] Inputting the aging characteristic parameter, the sound absorption decay amount, and the sound transmission gain amount into a preset device aging degree quantification model to obtain the device aging degree, wherein the device aging degree quantification model is constructed based on linear cumulative damage theory and gradient boosting decision tree;

[0010] According to the aging degree of the device, the remaining life of the noise reduction device is calculated based on Weibull distribution.

[0011] Furthermore, the step of collecting the operating data of the noise reduction device in the substation, performing feature extraction on the operating data, and obtaining aging feature parameters includes:

[0012] Collect vibration acceleration, temperature data, sound pressure data, gas flow data, dynamic stress time history data and acoustic emission signals of noise reduction devices in substations;

[0013] Performing frequency domain analysis on the vibration acceleration to obtain a main frequency energy ratio and a vibration main frequency offset, calculating a temperature change rate and a high temperature duration based on the temperature data, and calculating a coupling coefficient between the sound pressure data and the gas flow data based on a Pearson correlation coefficient;

[0014] Calculating the number of stress cycles according to the dynamic stress time history data, and performing statistics on the acoustic emission signals to obtain an acoustic emission event rate;

[0015] The main frequency energy proportion, the temperature change rate, the coupling coefficient, the number of stress cycles, the acoustic emission event rate, the high temperature duration and the vibration main frequency offset are used as aging characteristic parameters.

[0016] Furthermore, the step of performing cluster analysis on the aging characteristic parameters and obtaining the degradation stage and the corresponding sound absorption decay amount according to the clustering results includes:

[0017] The main frequency energy proportion, the temperature change rate and the coupling coefficient are used as initial characteristic parameters, and the principal component analysis method is used to perform feature dimensionality reduction on the initial characteristic parameters to obtain reduced-dimensional characteristic parameters, wherein the reduced-dimensional characteristic parameters include the interface debonding area and the material microporosity;

[0018] The aging characteristic parameters are updated according to the characteristic parameters after dimensionality reduction, and a preset clustering model is used to perform cluster analysis on the updated aging characteristic parameters to obtain the degradation stage of the noise reduction device and the corresponding sound absorption decay amount. The clustering model is constructed based on the K-means clustering algorithm.

[0019] Furthermore, the step of performing structural sound transmission simulation based on the finite element model to obtain simulated sound transmission loss, and obtaining a sound transmission gain according to the simulated sound transmission loss and a preset reference sound transmission loss includes:

[0020] Applying loads to the finite element model to obtain structural deformation and structural vibration velocity, wherein the loads include airflow pressure loads and vibration loads;

[0021] Taking the structural deformation and the structural vibration velocity as acoustic boundary conditions, the propagation law of the sound wave in the finite element model is calculated based on the frequency domain sound pressure wave equation to obtain the sound pressure distribution;

[0022] Calculating the difference between the incident sound pressure level and the transmitted sound pressure level according to the sound pressure distribution to obtain a simulated sound transmission loss;

[0023] The difference between the preset reference sound transmission loss and the simulated sound transmission loss is calculated to obtain the sound transmission gain.

[0024] Furthermore, the device aging degree quantification model includes an underlying physical model and a data-driven model, wherein the underlying physical model is constructed based on the linear cumulative damage theory, and the data-driven model is constructed based on the gradient boosting decision tree;

[0025] The input of the underlying physical model is the number of stress cycles, and the output of the underlying physical model is the basic aging degree;

[0026] The input of the data-driven model is the second aging characteristic parameter, the sound absorption decay amount and the sound transmission gain amount, and the output of the data-driven model is the correction factor;

[0027] The output of the device aging degree quantification model is the device aging degree, and the device aging degree is the product of the basic aging degree and the correction factor.

[0028] Furthermore, the basic aging degree is expressed by the following formula:

[0029]

[0030] Where D base Indicates the basic aging degree, n i Indicates the number of stress cycles at level i, N i represents the fatigue life of level i;

[0031] The following formula is used to express the aging degree of the device:

[0032]

[0033] Where Dg Indicates the aging degree of the device, and k is the correction factor.

[0034] Furthermore, the step of calculating the remaining life of the noise reduction device based on the Weibull distribution according to the aging degree of the device includes:

[0035] The aging degree of the device and the current working life of the noise reduction device are input into a pre-built life prediction model to obtain the remaining life of the noise reduction device. The life prediction model is constructed based on Weibull distribution, and the parameters of the life prediction model are estimated based on historical aging data.

[0036] Furthermore, the remaining life is expressed by the following formula:

[0037]

[0038] Where E represents the remaining life, t0 represents the current working years, η represents the scale parameter, β represents the shape parameter, represents the upper incomplete gamma function.

[0039] Furthermore, the step of calculating the remaining life of the noise reduction device based on the Weibull distribution according to the aging degree of the device includes:

[0040] According to the degradation stage of the noise reduction device, a corresponding stage life prediction model is selected from a pre-built life prediction model set;

[0041] Inputting the aging degree of the device and the current service life into the stage life prediction model to obtain the remaining life of the noise reduction device;

[0042] The life prediction model set includes multiple stage life prediction models, and the stage life prediction models are constructed based on Weibull distribution. The parameters of each stage life prediction model are estimated based on the corresponding stage historical aging data.

[0043] In a second aspect, the present invention provides a life prediction system for a noise reduction device for a substation, the system comprising:

[0044] A data processing module is used to collect operating data of the noise reduction device in the substation, perform feature extraction on the operating data, and obtain aging feature parameters;

[0045] a degradation quantification module, configured to perform cluster analysis on the aging characteristic parameters and obtain the degradation stage and the corresponding sound absorption degradation amount according to the clustering results;

[0046] a gain quantification module, configured to construct a finite element model of the noise reduction device according to the aging characteristic parameters, perform structure-borne sound transmission simulation based on the finite element model to obtain a simulated sound transmission loss, and obtain a sound transmission gain based on the simulated sound transmission loss and a preset reference sound transmission loss;

[0047] an aging quantification module, configured to input the aging characteristic parameters, the sound absorption degradation amount, and the sound transmission gain amount into a preset device aging degree quantification model to obtain the device aging degree, wherein the device aging degree quantification model is constructed based on linear cumulative damage theory and a gradient boosting decision tree;

[0048] The life prediction module is used to calculate the remaining life of the noise reduction device based on the Weibull distribution according to the aging degree of the device.

[0049] This invention provides a lifespan prediction method and system for noise reduction devices used in substations. By extracting aging features, the method improves the data quality of feature information. By quantifying the degradation of the noise reduction device's sound absorption and insulation functions and combining theoretical calculations with data-driven methods, the accuracy of device aging assessments is effectively improved. Furthermore, by using a reasonable distribution model to predict the remaining lifespan of the noise reduction device, an accurate assessment of the remaining lifespan can be achieved. This invention enables accurate aging assessment and lifespan prediction, providing reliable technical support for maintenance decisions for noise reduction devices. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 is a flow chart of a method for predicting the life of a noise reduction device for a substation according to an embodiment of the present invention;

[0051] Figure 2 This is a structural flow chart of a life prediction system for a noise reduction device used in a substation according to an embodiment of the present invention;

[0052] Reference numerals:

[0053] 10. Data processing module; 20. Decay quantification module; 30. Gain quantification module; 40. Aging quantification module; 50. Life prediction module. DETAILED DESCRIPTION

[0054] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0055] See also Figure 1 The first embodiment of the present invention provides a life prediction method for a noise reduction device for a substation, comprising steps S10 to S50:

[0056] Step S10, collecting operating data of the noise reduction device in the substation, performing feature extraction on the operating data, and obtaining aging feature parameters;

[0057] Step S20, performing cluster analysis on the aging characteristic parameters, and obtaining the degradation stage and the corresponding sound absorption decay amount according to the clustering results;

[0058] Step S30, constructing a finite element model of the noise reduction device according to the aging characteristic parameters, performing structure-borne sound transmission simulation based on the finite element model to obtain a simulated sound transmission loss, and obtaining a sound transmission gain based on the simulated sound transmission loss and a preset reference sound transmission loss;

[0059] Step S40: Inputting the aging characteristic parameter, the sound absorption decay amount, and the sound transmission gain amount into a preset device aging degree quantification model to obtain the device aging degree, wherein the device aging degree quantification model is constructed based on linear cumulative damage theory and a gradient boosting decision tree;

[0060] Step S50 : calculating the remaining life of the noise reduction device based on the Weibull distribution according to the aging degree of the device.

[0061] The present invention is aimed at the operation and maintenance management of noise reduction devices installed in substations, and provides a method for predicting the remaining service life of the noise reduction devices by analyzing the degree of aging. Before explaining the life prediction method, the noise reduction device in the present invention is briefly introduced. The noise reduction device in this embodiment adopts a conventional multi-layer composite structure design, which can be divided into a supporting skeleton, a sound-absorbing material layer, an interface bonding layer and a flow channel structure according to the functional structure. The noise reduction performance and status monitoring are achieved through the cooperation of various parts. Specifically, the support frame serves as the mechanical bearing core, providing mechanical support for the device, bearing external loads (such as airflow pressure and vibration impact), and constraining the spatial position of the sound-absorbing material layer. It is typically constructed of lightweight, high-strength metal or composite materials, with a grid-like or honeycomb-like hollow design. The sound-absorbing material layer, serving as the primary noise reduction component, absorbs sound energy through a porous structure or resonant units, converting it into heat or mechanical energy for dissipation. The interfacial bonding layer, made of a high-performance adhesive, serves as the structural connection hub, tightly connecting the sound-absorbing material layer to the support frame, transferring stress and preventing interfacial debonding, which can lead to sound leakage and reduce noise reduction performance. The flow channel structure serves as a fluid flow path, guiding airflow through the device. To facilitate degradation analysis, in this embodiment, multiple sensors are installed at corresponding locations on the noise reduction device. A multi-sensor array collects data on the device's operating status, providing data for subsequent degradation analysis. The noise reduction device is mainly installed on the key paths of noise propagation in the substation, such as the walls of the main transformer room, air inlets and outlets, and the outside of the guardrails. The specific structure and installation location can refer to the structure and location of conventional noise reduction devices in current substations, and no further restrictions are made here.

[0062] In this embodiment, a multi-sensor array is used to collect various operating data from the noise reduction device. Aging analysis and feature extraction are then performed on this operating data to obtain characteristic parameters used to characterize the device's aging state. Specifically, this embodiment utilizes multi-source sensors installed on the noise reduction device to collect data and obtain various operating data. This operating data includes vibration acceleration, temperature, sound pressure, gas flow, dynamic stress time history data, and acoustic emission signals. Because the device's mechanical structure may experience fatigue cracks, bolt loosening, or interface debonding under long-term vibration loads, these degradation behaviors directly affect the structural stiffness and natural frequency, thereby changing the vibration response. Therefore, vibration acceleration can be used to monitor the condition of the mechanical structure. Temperature data refers to the temperature distribution on and within the device. Since noise reduction devices are often integrated with heat-generating equipment (such as transformers and reactors), the surface and internal temperatures of the device directly affect material properties. Therefore, temperature data can be used to characterize the degree of thermal aging of the material. Sound pressure data refers to sound pressure data. Sound pressure is a direct indicator for evaluating the core functions (sound absorption / sound insulation) of the noise reduction device and can reflect the degree of noise reduction performance degradation. Gas flow data refers to the gas flow rate within the device. Gas flow rate is a comprehensive reflection of the material's permeability and structural sealing. Device aging, such as pore blockage in the sound-absorbing material or structural leakage, will affect the gas flow rate within the device. Therefore, gas flow data can be used to characterize the degree of material and structural aging. In addition, stress data is related to the fatigue life loss of the device, and acoustic emission signals are related to crack propagation in the device's structural framework. In other words, these operating data can indirectly characterize the aging status of the device. In this embodiment, the above-mentioned operating data can be collected by installing multiple-source sensors such as strain gauges, acoustic emission sensors, temperature sensors, infrared thermal imagers, and acceleration sensors at relevant positions of the noise reduction device.

[0063] The original operating data is then subjected to noise removal and correlation analysis to extract characteristic parameters that can further characterize the aging status of the device. The specific steps include:

[0064] Collect vibration acceleration, temperature data, sound pressure data, gas flow data, dynamic stress time history data and acoustic emission signals of noise reduction devices in substations;

[0065] Performing frequency domain analysis on the vibration acceleration to obtain a main frequency energy ratio and a vibration main frequency offset, calculating a temperature change rate and a high temperature duration based on the temperature data, and calculating a coupling coefficient between the sound pressure data and the gas flow data based on a Pearson correlation coefficient;

[0066] Calculating the number of stress cycles according to the dynamic stress time history data, and performing statistics on the acoustic emission signals to obtain an acoustic emission event rate;

[0067] The main frequency energy proportion, the temperature change rate, the coupling coefficient, the number of stress cycles, the acoustic emission event rate, the high temperature duration and the vibration main frequency offset are used as aging characteristic parameters.

[0068] In this embodiment, for the vibration acceleration, discretization processing is first performed to form a segmented vibration time series, and then each segment of the vibration time series is subjected to fast Fourier transform to convert the time domain signal into a frequency domain representation, and the main frequency is determined according to the power spectrum, and the main frequency energy and the total vibration energy are calculated. According to the ratio of the main frequency energy to the total vibration energy, the main frequency energy proportion is determined, and the difference between the current vibration main frequency and the preset healthy main frequency is calculated to obtain the vibration main frequency offset; for the temperature data, first, noise reduction processing is performed by Kalman filtering to obtain a noise-reduced temperature data sequence, and then the temperature change rate of adjacent sampling points is calculated according to the temperature data sequence, and the temperature change rate of the adjacent sampling points is statistically analyzed. The high temperature duration is obtained by calculating the cumulative time when the temperature exceeds the preset temperature; for the sound pressure data and gas flow rate data, the time series data of the sound pressure change and the gas flow rate are first time-aligned, and then the Pearson correlation coefficient between the two is calculated to obtain the coupling coefficient; for the number of stress cycles, the dynamic stress time history data is collected by strain gauges, and then the number of fatigue stress cycles of the skeleton material is counted based on the rain flow counting method; for the acoustic emission event rate, the acoustic emission signals generated by the expansion of skeleton microcracks are collected by acoustic emission sensors installed at key parts of the skeleton, and then the number of acoustic emission signal events generated by the expansion of skeleton microcracks per unit time is counted to obtain the acoustic emission event rate.

[0069] Then the main frequency energy proportion, temperature change rate, coupling coefficient, number of stress cycles, acoustic emission event rate, high temperature duration and vibration main frequency offset are used as aging characteristic parameters of the noise reduction device. Among these parameters, the number of stress cycles is directly correlated with fatigue life loss; the acoustic emission event rate is related to the crack growth rate; the duration of high temperature is related to the degree of material oxidation; and the shift in the main vibration frequency is related to the degree of reduction in the skeleton stiffness. The main frequency energy fraction reflects the energy distribution of the vibration signal, which is directly related to the skeleton stress. Therefore, stress concentration can lead to a decrease in structural stiffness, thereby changing the natural frequency. For example, when the skeleton stress increases, the natural frequency of the structure decreases, and the main frequency energy may shift from high frequency to low frequency. The temperature change rate is related to the material's microscopic porosity and the interfacial debonding area. This is because changes in the material's microstructure (such as thermal expansion and phase transition) lead to a lag in the temperature response. Therefore, the temperature change rate can indirectly reflect the stability of the microstructure. At the same time, the degradation of the interfacial bonding area (such as cracking of the bonding layer due to thermal stress) can also affect the heat transfer efficiency, thereby changing the temperature change rate. The coupling coefficient reflects the interaction between sound pressure and airflow. Interfacial debonding can cause flow channel deformation, thereby reducing the coupling coefficient. Therefore, the coupling coefficient is related to the interfacial debonding area. In other words, these parameters are all related to the degree of device aging.

[0070] One of the direct manifestations of the aging degree of the noise reduction device is the degradation of the sound absorption performance. The degradation of the sound absorption performance has stage characteristics. For example, in the early stage, it degrades slowly due to the clogging of the material micropores, in the middle stage, it degrades rapidly due to interface debonding, and in the final stage, it fails violently due to material fragmentation. At the same time, the aging characteristic parameters of the device are also different in different degradation stages. Therefore, by performing cluster analysis on the above-mentioned aging characteristic parameters, the current degradation stage of the device can be determined, thereby obtaining the sound absorption degradation corresponding to the degradation stage. Specifically, this embodiment uses a cluster analysis algorithm such as the K-means clustering algorithm to construct a clustering model, and uses historical data for training to obtain a trained clustering model. The extracted aging characteristic parameters are then input into the clustering model to output the current degradation stage of the noise reduction device and the corresponding sound absorption degradation.

[0071] In order to improve the accuracy of the clustering results, in a preferred embodiment, the present invention further processes the characteristic parameters before performing cluster analysis on the aging characteristic parameters. The specific steps include:

[0072] The main frequency energy proportion, the temperature change rate and the coupling coefficient are used as initial characteristic parameters, and the principal component analysis method is used to perform feature dimensionality reduction on the initial characteristic parameters to obtain reduced-dimensional characteristic parameters, wherein the reduced-dimensional characteristic parameters include the interface debonding area and the material microporosity;

[0073] The aging characteristic parameters are updated according to the characteristic parameters after dimensionality reduction, and a preset clustering model is used to perform cluster analysis on the updated aging characteristic parameters to obtain the degradation stage of the noise reduction device and the corresponding sound absorption decay amount. The clustering model is constructed based on the K-means clustering algorithm.

[0074] In this embodiment, the relationship between the parameters in the aging characteristic parameters and device aging is indirectly inferred rather than directly characterized. For example, the main frequency energy ratio, temperature change rate, and coupling coefficient indirectly reflect the degree of aging by characterizing the device's material microporosity and interface debonding area. To focus the data more closely on the core mechanism of device degradation and provide accurate and mechanistically clear input for subsequent aging analysis, this embodiment uses principal component analysis to reduce the dimensionality of these indirect parameters. This indirect data is mapped to a low-dimensional space using a projection matrix, converting them into target features that more directly reflect the device's degradation state. This results in parameters such as interface debonding area and material microporosity that directly characterize the degree of device aging. These reduced-dimensional parameters are then used to update the original aging characteristic parameters. The updated aging characteristic parameters include interface debonding area, material microporosity, number of stress cycles, acoustic emission event rate, high-temperature duration, and vibration main frequency offset. These parameters are then input into a clustering model for cluster analysis to determine the device's degradation stage and, consequently, the corresponding sound absorption decay. In this embodiment, the sound absorption degradation is a quantified value of the degradation of the sound absorption performance of the noise reduction device. The degradation stage and corresponding sound absorption degradation can be obtained through statistical analysis of historical degradation data. It should be noted that the cluster analysis in this embodiment may also employ other clustering algorithms. The specific clustering analysis steps can refer to the calculation steps of the clustering algorithm employed and are not limited herein.

[0075] Aging of noise reduction devices can lead to functional degradation. In addition to degradation of the sound absorption function, it also causes degradation of the sound insulation function. In this embodiment, degradation of the sound absorption function is quantified using sound absorption decay, while degradation of the sound insulation function is quantified using sound transmission gain. Specifically, the sound absorption function primarily refers to the ability of the sound-absorbing material (e.g., porous sound-absorbing cotton, resonant sound-absorbing structure) to absorb sound energy, while the sound insulation function primarily refers to the ability of the sound insulation structure (e.g., skeleton, partition) to isolate the transmission of sound waves. Therefore, sound absorption decay and sound transmission gain represent two different functional degradations with different physical meanings. Their primary causes correspond to the degradation of material properties (sound-absorbing material) and structural integrity (skeletal structure), respectively. These are two independent dimensions of the device's acoustic function. To quantify multi-dimensional degradation, in this embodiment, in addition to calculating the sound absorption decay, the sound transmission gain is also quantified through sound transmission simulation using a finite element model.

[0076] In this embodiment, a finite element model is first constructed based on the geometric shape, material parameters and aging characteristic parameters of the noise reduction device. Specifically, based on the architecture of the noise reduction device, the finite element model can be divided into a structural module and an acoustic module. The structural module is mainly a representation of the architecture of the noise reduction device, including a supporting skeleton, a sound-absorbing material layer, an interface bonding layer and a flow channel structure. The acoustic module is a representation of the sound field of the noise reduction device, including the propagation of sound waves inside the device and the far-field radiation of the external sound field. The degradation state of the model is modeled based on the aging characteristic parameters. The specific modeling steps can refer to the conventional modeling steps of the finite element model, which will not be repeated here. For the constructed finite element model, this embodiment performs structural sound transmission simulation by applying loads and calculates the sound transmission gain. The specific steps include:

[0077] Applying loads to the finite element model to obtain structural deformation and structural vibration velocity, wherein the loads include airflow pressure loads and vibration loads;

[0078] Taking the structural deformation and the structural vibration velocity as acoustic boundary conditions, the propagation law of the sound wave in the finite element model is calculated based on the frequency domain sound pressure wave equation to obtain the sound pressure distribution;

[0079] Calculating the difference between the incident sound pressure level and the transmitted sound pressure level according to the sound pressure distribution to obtain a simulated sound transmission loss;

[0080] The difference between the preset reference sound transmission loss and the simulated sound transmission loss is calculated to obtain the sound transmission gain.

[0081] In this embodiment, the purpose of constructing a finite element model and performing sound transmission simulation is to quantify the impact of aging on the device's sound insulation performance. The device's sound insulation performance is directly related to its structural integrity and is also affected by material properties. During the aging process, the skeleton deforms due to airflow pressure and vibration loads, while the interfacial bonding layer gradually debonds due to stress concentration. This degradation alters the device's "sound propagation path" (e.g., creating new leakage gaps) and "vibration-acoustic coupling characteristics" (e.g., debonding leading to more intense structural vibration), ultimately leading to a decrease in sound transmission loss, or in other words, a sound transmission gain. Therefore, the core goal of the simulation in this embodiment is to quantify the sound insulation degradation process through multi-physics field coupling calculations.

[0082] Specifically, airflow pressure and vibration loads are first applied to the structural module of the finite element model. The airflow pressure simulates the static pressure exerted by internal or external airflow on the skeleton during device operation, while the vibration load simulates the dynamic excitation of the skeleton by mechanical vibrations, such as those caused by motor operation. The structural deformation of the model is then calculated, including the deformation of the skeleton and the expansion of the interfacial bonding and detachment area. For example, the offset of the skeleton grid is calculated based on solid mechanics, and the proportion of the interfacial bonding and detachment area is calculated based on the cohesive force model. Simultaneously, the vibration velocity of the device structure after the load is applied is determined through structural analysis. The structural deformation and vibration velocity are then used as acoustic boundary conditions to solve the propagation law of sound waves in the aging structure and obtain the sound pressure distribution. The reason for applying loads to the finite element model for sound transmission simulation is that the essence of aging is the degradation of performance over time under the action of load, rather than static parameter attenuation. In the actual aging process, load is the driving factor of degradation, which in turn causes the decline of acoustic performance, namely, sound transmission loss. If only the initial degradation parameters are used for simulation and the dynamic degradation under the action of load is ignored, the simulation results will not accurately reflect the aging process in actual operation. The reason for using the finite element model to simulate the quantification of sound insulation degradation is that the main factor of sound insulation performance is the integrity of the device skeleton, and the device structure can be accurately modeled by the finite element and combined with characteristic parameters to characterize the aging state. Through the accurate modeling of the finite element model, the results of the finite element simulation calculation method will be more accurate than other methods.

[0083] This embodiment uses the frequency-domain sound pressure fluctuation equation (i.e., the Helmholtz equation) as the propagation equation for acoustic wave aging structures. Structural vibrations caused by airflow pressure and vibration loads excite the surrounding air to generate sound waves, while structural deformations (such as grille offset) change the propagation path and reflection / transmission characteristics of the sound waves. In the frequency-domain sound pressure fluctuation equation, the structural vibration velocity directly affects the sound source term of the sound wave equation through boundary conditions, and structural deformation changes the geometry of the acoustic calculation domain (such as the channel cross-sectional area and boundary shape), thereby affecting the spatial distribution of the sound pressure.

[0084] Based on the above description, the frequency domain sound pressure fluctuation equation can be expressed as:

[0085]

[0086] Where p represents the sound pressure, represents the air density, c represents the speed of sound, represents the angular frequency, which is determined by the frequency of the incident sound wave, v represents the structural vibration velocity, n represents the interface normal vector, ▽ represents the Hamiltonian operator, and i represents the imaginary unit.

[0087] In the above propagation formula, the first term on the left reflects the diffusion of sound pressure, the second term on the left reflects the inertia of the sound wave, and the term on the right is the source of the excitation of the sound wave by the structural vibration. For this formula, the equation is discretized using the finite element method, and the structural vibration velocity obtained from structural analysis is used as the boundary condition (i.e., the sound source term). The spatial distribution of the sound pressure is solved by meshing the deformed structural geometry. Based on this spatial distribution of sound pressure, the inlet and outlet sound pressure levels of the device—the incident sound pressure level and the transmitted sound pressure level—are then extracted. The transmitted sound pressure level is subtracted from the incident sound pressure level to obtain the simulated sound transmission loss. Finally, the calculated simulated sound transmission loss is subtracted from the baseline sound transmission loss for the device in its healthy state to obtain the sound transmission gain.

[0088] In this embodiment, by driving the finite element coupling simulation of multi-physics fields with aging parameters, a quantitative analysis of structural aging and sound propagation changes is achieved, thereby providing an explainable quantitative basis for the degradation of the noise reduction performance of the device.

[0089] After quantitatively analyzing the sound absorption and sound insulation capabilities of the noise reduction device, this embodiment pre-constructs a quantitative model of the device aging degree, and inputs these aging characteristic parameters together with the sound absorption decay and sound transmission gain into the quantitative model of the device aging degree to predict the device aging degree. At the same time, in order to reduce the amount of calculation, in a preferred embodiment, a gain threshold can be set. When the sound transmission gain is greater than the gain threshold, the device aging degree prediction is performed.

[0090] In this embodiment, the device aging degree quantification model adopts a fusion model of the gradient boosting decision tree and the physical empirical formula. Its model structure includes an underlying physical model and a data-driven model. The underlying physical model is constructed based on Miner's linear cumulative damage theory and is used to define the basic aging degree. Its formula is expressed as follows:

[0091]

[0092] Where D base Indicates the basic aging degree, n i Indicates the number of stress cycles at level i, N i represents the fatigue life of the i-th level, where the number of stress cycles is obtained by statistically analyzing the collected dynamic stress time history data using the rain flow counting method, that is, it can be extracted from the aging characteristic parameters. The fatigue life is determined by fatigue tests (such as SN curve tests). After the number of stress cycles is input into the underlying physical model, the corresponding fatigue life is extracted according to the number of stress cycles, thereby calculating the basic aging degree.

[0093] The data-driven model is based on a gradient boosting decision tree. Its inputs are aging characteristic parameters, sound absorption degradation, and sound transmission gain. Its output is a correction factor, which adjusts the baseline aging level to accommodate multi-source degradation mechanisms. The training steps for the data-driven model can be compared to the conventional training steps for gradient boosting decision trees and will not be detailed here.

[0094] Finally, the correction factor is multiplied by the basic aging degree to obtain the aging degree of the noise reduction device, which is expressed as follows:

[0095]

[0096] Where D g Indicates the aging degree of the device, and k is the correction factor.

[0097] This embodiment combines theoretical calculation with data-driven approach, so that the model can not only utilize the physical significance of Miner theory, but also correct its limitations through multi-source data, thereby effectively improving the accuracy of aging assessment.

[0098] After obtaining the aging degree of the device, this embodiment uses the Weibull distribution to calculate the remaining life of the noise reduction device. The Weibull distribution is a nonlinear model that can flexibly describe the time-varying nature of the aging rate through shape parameters. It is highly consistent with the actual multi-stage aging process of the noise reduction device. Therefore, this embodiment uses the Weibull distribution to construct a life prediction model. Specifically, the cumulative distribution function of the Weibull distribution is used as the aging degree:

[0099]

[0100] Where D(t) represents the probability of device failure before time t, which can be understood as the degree of aging at time t, η represents the scale parameter, and β represents the shape parameter.

[0101] The probability that the device has not failed at time t, that is, the survival function S(t), can be expressed as:

[0102]

[0103] The remaining life is defined as the time that the device can still work normally in the future under the condition that it has survived to time t0. The survival probability of the remaining life is the conditional survival function :

[0104]

[0105] Where t0 represents the survival time, which can be understood as the current working life of the device, and △t represents the future time period.

[0106] The average remaining life is the data expectation of the remaining life, reflecting the average remaining time of the device at t0, which can be expressed as:

[0107]

[0108] The cumulative distribution function is calculated by the incomplete gamma function. By replacing variables, the lower limit of the integral is equivalent to (t0+△t) / η, and the remaining life E is simplified to:

[0109]

[0110] Where E represents the remaining life, t0 represents the current working years, η represents the scale parameter, β represents the shape parameter, represents the upper incomplete gamma function.

[0111] The scale parameter and shape parameter in the remaining life formula need to be estimated through historical aging data. Specifically, through accelerated aging experiments or actual operation monitoring, multiple sets of time and aging degree data are obtained, which are used as the original data sets of t and D(t) in the cumulative distribution function of the Weibull distribution. The cumulative distribution function is then linearized through double logarithmic changes, and the data is fitted by maximum likelihood estimation or least squares method based on the original data set to solve for the scale parameter and shape parameter.

[0112] Through the above steps, a life prediction model based on the remaining life formula can be constructed, and then the calculated device aging degree and the current working life of the noise reduction device are input into the model to obtain the remaining life of the noise reduction device.

[0113] In a preferred embodiment, the step of calculating the remaining life of the noise reduction device based on the Weibull distribution includes:

[0114] According to the degradation stage of the noise reduction device, a corresponding stage life prediction model is selected from a pre-built life prediction model set;

[0115] Inputting the aging degree of the device and the current service life into the stage life prediction model to obtain the remaining life of the noise reduction device;

[0116] The life prediction model set includes multiple stage life prediction models, and the stage life prediction models are constructed based on Weibull distribution. The parameters of each stage life prediction model are estimated based on the corresponding stage historical aging data.

[0117] The main difference between the life prediction model in this embodiment and the life prediction model in the previous embodiment is that the life prediction model in this embodiment includes multiple stage life prediction models. Since the aging of the noise reduction device is time-varying, that is, it has the characteristics of slowness in the early stage, acceleration in the middle stage, and severe degradation in the late stage, in order to improve the accuracy of life prediction, this embodiment divides the original data group according to the degradation stage, and then fits the corresponding shape parameters and scale parameters according to different degradation stages to obtain multiple groups of parameters, so as to obtain stage life prediction models corresponding to different degradation stages. The degradation stage used in the parameter fitting calculation is consistent with the degradation stage of the cluster analysis stage. Therefore, after determining the degradation stage and calculating the aging degree of the device by clustering analysis of the aging characteristic parameters, the stage life prediction model corresponding to the degradation stage can be selected from the life prediction model set according to the degradation stage, and then the aging degree of the device and the current working life are input into the corresponding stage life prediction model to obtain the remaining life of the noise reduction device.

[0118] This embodiment provides a life prediction method for a noise reduction device for a substation. By extracting aging features, the data quality of feature information is improved. Cluster analysis and multi-physical layer coupling analysis are used to quantify the degree of degradation of the noise reduction device's sound absorption and sound insulation functions. Combining aging feature parameters and employing a combination of theoretical calculation and data-driven methods effectively improves the accuracy of the device aging assessment. Furthermore, a reasonable distribution model is used to predict the remaining life of the noise reduction device, achieving an accurate assessment of the remaining life. Through feature extraction, coupling analysis, model correction, and distributed calculation, the present invention achieves precise mapping from multi-source signals to aging states. Accurate aging assessment and life prediction provide reliable technical support for maintenance decisions for noise reduction devices.

[0119] See also Figure 2 Based on the same inventive concept, a second embodiment of the present invention provides a life prediction system for a noise reduction device for a substation, comprising:

[0120] The data processing module 10 is used to collect the operating data of the noise reduction device in the substation, perform feature extraction on the operating data, and obtain aging feature parameters;

[0121] The degradation quantification module 20 is used to perform cluster analysis on the aging characteristic parameters and obtain the degradation stage and the corresponding sound absorption degradation amount according to the clustering results;

[0122] a gain quantification module 30 for constructing a finite element model of the noise reduction device according to the aging characteristic parameters, performing structure-borne sound transmission simulation based on the finite element model to obtain a simulated sound transmission loss, and obtaining a sound transmission gain based on the simulated sound transmission loss and a preset reference sound transmission loss;

[0123] an aging quantification module 40 for inputting the aging characteristic parameters, the sound absorption degradation amount, and the sound transmission gain amount into a preset device aging degree quantification model to obtain the device aging degree, wherein the device aging degree quantification model is constructed based on linear cumulative damage theory and a gradient boosting decision tree;

[0124] The life prediction module 50 is used to calculate the remaining life of the noise reduction device based on the Weibull distribution according to the aging degree of the device.

[0125] The technical features and effects of the life prediction system for noise reduction devices for substations proposed in the embodiments of the present invention are the same as those of the method proposed in the embodiments of the present invention and are not further described here. Each module in the above-mentioned life prediction system for noise reduction devices for substations can be implemented in whole or in part through software, hardware, or a combination thereof. Each of the above-mentioned modules can be embedded in or independent of the processor in the computer device in hardware form, or can be stored in the memory of the computer device in software form, so that the processor can call and execute the corresponding operations of each of the above modules.

[0126] In summary, an embodiment of the present invention proposes a life prediction method and system for a noise reduction device for a substation. The method collects operating data of the noise reduction device in the substation, extracts features from the operating data, and obtains aging characteristic parameters; performs cluster analysis on the aging characteristic parameters, and obtains the degradation stage and the corresponding sound absorption decay amount based on the clustering results; constructs a finite element model of the noise reduction device according to the aging characteristic parameters, performs structural sound transmission simulation based on the finite element model to obtain simulated sound transmission loss, and obtains the sound transmission gain according to the simulated sound transmission loss and a preset benchmark sound transmission loss; inputs the aging characteristic parameters, the sound absorption decay amount, and the sound transmission gain amount into a preset device aging degree quantification model to obtain the device aging degree, and the device aging degree quantification model is constructed based on the linear cumulative damage theory and the gradient boosting decision tree; and calculates the remaining life of the noise reduction device based on the Weibull distribution according to the device aging degree. This invention improves the data quality of feature information through aging feature extraction. It quantifies the degree of degradation of the noise reduction device's sound absorption and insulation functions through cluster analysis and multi-physical layer coupling analysis. By combining aging feature parameters and employing a combination of theoretical calculation and data-driven methods, it effectively improves the accuracy of device aging assessments. It also predicts the remaining life of the noise reduction device through a reasonable distribution model, enabling accurate assessment of the remaining life. Through feature extraction, coupling analysis, model modification, and distributed calculation, this invention achieves precise mapping from multi-source signals to aging states. Accurate aging assessment and life prediction provide reliable technical support for maintenance decisions for noise reduction devices.

[0127] Each embodiment in this specification is described in a progressive manner, and the same or similar parts of each embodiment can be directly referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment. It should be noted that the various technical features of the above embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the various technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0128] The above-described embodiments merely represent several preferred implementations of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the patent. It should be noted that a person skilled in the art could make several improvements and substitutions without departing from the technical principles of the present invention, and these improvements and substitutions should also be considered within the scope of protection of the present application. Therefore, the scope of protection of the present patent application shall be based on the scope of protection of the claims.

Claims

1. A life prediction method for a noise reduction device for a substation, characterized in that: include: Collecting operating data of a noise reduction device in a substation, performing feature extraction on the operating data, and obtaining aging feature parameters; Performing cluster analysis on the aging characteristic parameters, and obtaining the degradation stage and the corresponding sound absorption decay amount according to the clustering results; constructing a finite element model of the noise reduction device according to the aging characteristic parameters, performing a structure-borne sound transmission simulation based on the finite element model to obtain a simulated sound transmission loss, and obtaining a sound transmission gain according to the simulated sound transmission loss and a preset reference sound transmission loss; Inputting the aging characteristic parameter, the sound absorption decay amount, and the sound transmission gain amount into a preset device aging degree quantification model to obtain the device aging degree, wherein the device aging degree quantification model is constructed based on linear cumulative damage theory and gradient boosting decision tree; According to the aging degree of the device, the remaining life of the noise reduction device is calculated based on Weibull distribution.

2. The life prediction method for a noise reduction device for a substation according to claim 1, characterized in that: The step of collecting the operating data of the noise reduction device in the substation, performing feature extraction on the operating data, and obtaining aging feature parameters includes: Collect vibration acceleration, temperature data, sound pressure data, gas flow data, dynamic stress time history data and acoustic emission signals of noise reduction devices in substations; Performing frequency domain analysis on the vibration acceleration to obtain a main frequency energy ratio and a vibration main frequency offset, calculating a temperature change rate and a high temperature duration based on the temperature data, and calculating a coupling coefficient between the sound pressure data and the gas flow data based on a Pearson correlation coefficient; Calculating the number of stress cycles according to the dynamic stress time history data, and performing statistics on the acoustic emission signals to obtain an acoustic emission event rate; The main frequency energy proportion, the temperature change rate, the coupling coefficient, the number of stress cycles, the acoustic emission event rate, the high temperature duration and the vibration main frequency offset are used as aging characteristic parameters.

3. The life prediction method for a noise reduction device for a substation according to claim 2, characterized in that: The step of performing cluster analysis on the aging characteristic parameters and obtaining the degradation stage and the corresponding sound absorption decay amount according to the clustering results includes: The main frequency energy proportion, the temperature change rate and the coupling coefficient are used as initial characteristic parameters, and the principal component analysis method is used to perform feature dimensionality reduction on the initial characteristic parameters to obtain reduced-dimensional characteristic parameters, wherein the reduced-dimensional characteristic parameters include the interface debonding area and the material microporosity; The aging characteristic parameters are updated according to the characteristic parameters after dimensionality reduction, and a preset clustering model is used to perform cluster analysis on the updated aging characteristic parameters to obtain the degradation stage of the noise reduction device and the corresponding sound absorption decay amount. The clustering model is constructed based on the K-means clustering algorithm.

4. The life prediction method for a noise reduction device for a substation according to claim 1, characterized in that: The steps of performing structural sound transmission simulation based on the finite element model to obtain simulated sound transmission loss, and obtaining a sound transmission gain according to the simulated sound transmission loss and a preset reference sound transmission loss include: Applying loads to the finite element model to obtain structural deformation and structural vibration velocity, wherein the loads include airflow pressure loads and vibration loads; Taking the structural deformation and the structural vibration velocity as acoustic boundary conditions, the propagation law of the sound wave in the finite element model is calculated based on the frequency domain sound pressure wave equation to obtain the sound pressure distribution; Calculating the difference between the incident sound pressure level and the transmitted sound pressure level according to the sound pressure distribution to obtain a simulated sound transmission loss; The difference between the preset reference sound transmission loss and the simulated sound transmission loss is calculated to obtain the sound transmission gain.

5. The life prediction method for a noise reduction device for a substation according to claim 2, characterized in that: The device aging degree quantification model includes an underlying physical model and a data-driven model, wherein the underlying physical model is constructed based on the linear cumulative damage theory, and the data-driven model is constructed based on the gradient boosting decision tree; The input of the underlying physical model is the number of stress cycles, and the output of the underlying physical model is the basic aging degree; The input of the data-driven model is the aging characteristic parameter, the sound absorption decay amount and the sound transmission gain amount, and the output of the data-driven model is the correction factor; The output of the device aging degree quantification model is the device aging degree, and the device aging degree is the product of the basic aging degree and the correction factor.

6. The life prediction method for a noise reduction device for a substation according to claim 5, characterized in that: The basic aging degree is expressed by the following formula: Where D base Indicates the basic aging degree, n i Indicates the number of stress cycles at level i, N i represents the fatigue life of level i; The following formula is used to express the aging degree of the device: Where D g Indicates the aging degree of the device, and k is the correction factor.

7. The life prediction method for a noise reduction device for a substation according to claim 1, characterized in that: The step of calculating the remaining life of the noise reduction device based on the Weibull distribution according to the aging degree of the device includes: The aging degree of the device and the current working life of the noise reduction device are input into a pre-built life prediction model to obtain the remaining life of the noise reduction device. The life prediction model is constructed based on Weibull distribution, and the parameters of the life prediction model are estimated based on historical aging data.

8. The life prediction method for a noise reduction device for a substation according to claim 7, characterized in that: The remaining life is expressed using the following formula: Where E represents the remaining life, t0 represents the current working years, η represents the scale parameter, β represents the shape parameter, represents the upper incomplete gamma function.

9. The life prediction method for a noise reduction device for a substation according to claim 7, characterized in that: The step of calculating the remaining life of the noise reduction device based on the Weibull distribution according to the aging degree of the device includes: According to the degradation stage of the noise reduction device, a corresponding stage life prediction model is selected from a pre-built life prediction model set; Inputting the aging degree of the device and the current service life into the stage life prediction model to obtain the remaining life of the noise reduction device; The life prediction model set includes multiple stage life prediction models, and the stage life prediction models are constructed based on Weibull distribution. The parameters of each stage life prediction model are estimated based on the corresponding stage historical aging data.

10. A life prediction system for a noise reduction device used in a substation, characterized in that: include: A data processing module is used to collect operating data of the noise reduction device in the substation, perform feature extraction on the operating data, and obtain aging feature parameters; a degradation quantification module, configured to perform cluster analysis on the aging characteristic parameters and obtain the degradation stage and the corresponding sound absorption degradation amount according to the clustering results; a gain quantification module, configured to construct a finite element model of the noise reduction device according to the aging characteristic parameters, perform structure-borne sound transmission simulation based on the finite element model to obtain a simulated sound transmission loss, and obtain a sound transmission gain based on the simulated sound transmission loss and a preset reference sound transmission loss; an aging quantification module, configured to input the aging characteristic parameters, the sound absorption degradation amount, and the sound transmission gain amount into a preset device aging degree quantification model to obtain the device aging degree, wherein the device aging degree quantification model is constructed based on linear cumulative damage theory and a gradient boosting decision tree; The life prediction module is used to calculate the remaining life of the noise reduction device based on the Weibull distribution according to the aging degree of the device.

Images